Académique Documents
Professionnel Documents
Culture Documents
/
y
Note: The material properties used in this figure are obtained from AA (2000a
and 2000b) with an assumption that the strain at the ultimate stress is
approximately half of the minimum percent elongation. The points
corresponding to the yield and ultimate stresses are shown with solid circles. F
y
denotes the yield stress and
y
denotes the strain corresponding to the yield
stress.
Figure 1.1 Normalized uniaxial tensile stress-strain curves of extruded aluminum
alloys generated by Ramberg-Osgood equation
One type of the standard extruded shapes available in the Aluminum Design
Manual by the Aluminum Association (2000b) is cross-sections with tapered
thickness, which were to facilitate the rolling process when the extrusion process was
not common or available. This type of section is also common in steel. The buckling
behavior of tapered thickness component elements is quite different from that of
3
uniform thickness elements. However, neither the Steel Specification by the American
Institute of Steel Construction (AISC, 1998) nor the AA Specification (2000a)
provides a way to account for the slope of tapered thickness elements. Instead,
treatment of a tapered component element as a uniform element using an averaged
thickness is implied.
The study presented in this report is aimed at improving the current AA
Specification (2000a) so that the aforementioned problems can be properly resolved.
All the investigations in this study are based on laterally supported members or
members that do not buckle laterally under flexural loading.
In Chapter 2, the ultimate-plastic capacity is studied. To begin with, important
terminologies used in this study are defined. An analytical closed-form equation of the
ultimate shape factor is derived for rectangular web elements. Using this equation, the
ultimate shape factor implicit in the AA Specification (2000a) is examined for almost
a thousand available alloy-temper combinations. In addition, a parametric study is
conducted to account for the inelastic buckling behavior of component elements in
compression. Additional studies are conducted on the approaches in the AA
Specification (2000a) that cover the limit state stresses for all component elements to
compute the member moment capacity. These will be referred to as the moment
capacity evaluation approaches. In addition, a more general approach than those in
the present specification is developed. Parametric studies using finite element analyses
as well as physical tests are conducted to validate the proposed approaches in this
chapter. Finally, to maintain a certain factor of safety on yield strength, the use of a
specified percentage of the ultimate strength of a member with two safeguards is
suggested.
In Chapter 3, the behavior of sections with tapered thickness is investigated.
To understand the buckling behavior of a tapered thickness plate, stiffness matrices
are derived for use in an available finite strip program. Using the program, the
4
buckling coefficients for tapered plates are computed. Based on the plate buckling
coefficients, the current specification equations are modified for the tapered thickness
component element. To validate the modified equations, a parametric study of I-
sections with tapered flanges with a wide variety of slenderness is performed and
physical flexural tests are conducted.
In Chapter 4, a general design approach is developed for complex extrusions,
for which simply-supported boundary idealization is not applicable or appropriate.
This approach relies on a numerical buckling analysis tool. In addition, the rigorous
analytic expression of the ultimate shape factor developed for symmetric sections in
Chapter 2 is extended to the ultimate shape factor for unsymmetric sections.
Additional moment capacity evaluation approaches are developed for cross-sections
with a neutral axis that is not at mid-depth. Parametric studies of five different kinds
of cross-sections are conducted using finite element analysis to validate the
approaches developed in this chapter. Some additional physical tests are also
conducted for further validation.
In Chapter 5, major conclusions of the study are provided. In the Appendix,
step-by-step design examples are provided. In addition, detailed cross-sectional
dimensions and numerical results of most of the parametric studies are tabulated for
future reference. Moreover, uniaxial tension test results and initial geometric
imperfection measurements conducted in this study are summarized.
The studies in this report refer to valuable studies by a number of researchers.
First, for understanding the behavior of aluminum structures, textbooks by Kissell and
Ferry (1995), Sharp (1993), Mazzolani (1985) and Mazzolani (1995) have been very
useful. The first two are based on the AA Specification (2000a); the others are related
to the European specifications, such as Eurocode 9 (1997). Templin et al. (1938),
Stowell (1948), Stowell (1950), Bleich (1952), Hill and Clark (1955), Anderson and
Anderson (1956), Alcoa (1958), Clark and Rolf (1966), Sharp (1966), Jombock and
5
Clark (1968) and Sooi and Pekz (1993) have provided a significant portion of bases
for the AA Specification (2000a) with respect to the compressive strength of
component elements. Plecher (2000) has concentrated on evaluating existing
aluminum bridges in the United States.
The ultimate-plastic capacity of aluminum structures studied in Chapter 2 has
also been studied by Sharp (in Gaylord and Gaylord, 1979), Mazzolani (1985),
Mazzolani and Piluso (1997), Faella et al. (2000), De Matteis et al. (2001), and a
number of other researchers. For the development of the stiffness matrices for the
tapered thickness plate element in Chapter 3, the studies by Kobayashi et al. (1990),
Mizusawa (1993), Ohga et al. (1995), Cheung (1976), McGuire et al. (2000), Schafer
(1997), and Huebner et al. (1995) have been investigated. Attempts to incorporate
numerical buckling analyses into the specification approaches, as is done in Chapter 4,
have also been made by Schafer and Pekz (1998), and Mennink (2002).
6
2. SYMMETRIC CROSS-SECTIONS WITH UNIFORM
THICKNESS COMPONENT ELEMENTS
Most standard cross-sections listed in AA (2000b) are symmetric with respect
to at least one principal bending axis without any edge or intermediate stiffeners. The
cross-sectional shapes that fall into this category are exemplified in Figure 2.1. The
shapes are composed of equal flanges at the extreme fibers and of one or more web
elements.
Figure 2.1 Examples of standard simple extrusions
Once such a member is supported against lateral movement, the moment
capacity of the member mainly depends on the component elements. Specifications
customarily assume junctions between the component elements as simply supported
hinges, e.g., the American Institute of Steel Construction (AISC, 1998), American
Iron and Steel Institute (AISI, 1996), and Aluminum Association (AA, 2000a). As
such, each component element can be treated as an independent plate element under
corresponding boundary and loading conditions, of which the buckling behavior has
been shown, e.g., see Gerard and Becker (1957). Such cross-sections are referred to as
simple extrusions hereafter. Although extensive studies have been performed to
evaluate the precise flexural capacity of aluminum simple extrusions, there is still
room for improvement as shown in this study.
7
2.1 Introduction to Shape Factor and Limit State
For a steel member, the ratio of the plastic moment capacity (M
p
, Figure 2.2d)
to the yield moment capacity (M
y
= F
y
S, Figure 2.2b) is generally called the shape
factor. The computation of the shape factor is straightforward, since the stress
distribution is assumed to be rigid-plastic for the plastic moment capacity and linear-
elastic for the yield moment capacity. Since the stress distributions are linear, the
shape factor is dependent on the geometric shape and not the material properties. For
example, the shape factor for a solid rectangular cross-section is 1.5, regardless of the
material properties.
On the other hand, the shape factor of an aluminum member is more
complicated: the stress distribution cannot be assumed to be rigid-plastic because the
stress-strain curve for aluminum does not have a constant yielding range, and strain
hardening occurs without a clearly defined yield point. So for aluminum, when the
stresses at both extreme fibers are at the ultimate stress, as shown in Figure 2.2e, this
stress distribution is defined as the limit state of the ultimate stress, and hereafter will
be referred to as the ultimate limit state.
Figure 2.2 Stress distributions of a rectangular aluminum section
The corresponding moment capacity is obtained through the integration of the non-
linear stress distribution, which is defined as the moment capacity at the ultimate limit
state and hereafter will be referred to as the ultimate moment capacity (M
u
):
/ 2
/ 2
h
u
h
M b f y dy
=
(1)
8
The ratio of the ultimate moment capacity (M
u
) to the yield moment (M
y
) is defined as
the shape factor for the ultimate limit state and will be referred to as the ultimate shape
factor (
u
). This ultimate shape factor is dependent on the geometric shape and
material properties including strain hardening.
In addition to the ultimate shape factor, another shape factor is used for
aluminum members. When the stresses at both extreme fibers are at the yield stress as
shown in Figure 2.2c, the moment capacity calculated through integration of the stress
distribution is defined as the moment capacity at the yield limit state. The moment
capacity at this state is significantly larger than the yield moment capacity (M
y
),
because the proportional limit of the stress-strain curve is appreciably smaller than the
yield stress based on the 0.2% offset method. The ratio of the moment capacity at the
yield limit state to the yield moment capacity (M
y
) is defined as the shape factor for
the yield limit state and will be referred to as the yield shape factor (
y
).
The shape factors for both the ultimate and yield limit states can be simplified
into Equation (2):
y
M
M
= (2)
where M is the moment capacity obtained from integrating the stress distribution at
either the yield or ultimate limit state. The shape factors can also be expressed in
terms of the stresses: the combination of Equation (2) and F
p
M/S results in
p
y
F
F
= (3)
where F
p
denotes the limit state stress. The limit state stress is a convenient measure
of the capacity of a component element, since the corresponding moment capacity is
simply obtained from the multiplication of the limit state stress by the elastic section
modulus. As a result of this process, the non-linear stress distribution of an aluminum
alloy is linearized as shown by the dashed lines in Figure 2.2c and Figure 2.2e. For a
9
component element under compression, the limit state stress varies according to the
width-to-thickness ratio of the cross-section. However, for a component element under
tension, the limit state stress is constant for each limit state. Details regarding the limit
state stress can be found in Section 2.2.
The results of numerical analysis and experimental studies shown in Table 2.1
present a value of 1.3 for the yield shape factor (
y
) of solid rectangular aluminum
flexural members. However, it is noted that the ultimate shape factor (
u
) proposed by
Sharp (1993) is considerably larger than the yield shape factor because the ultimate
stress, F
u
, is always larger than the yield stress, F
y
.
Table 2.1 Shape factors for solid rectangular aluminum flexural members
yield shape factor
by
Clark and Rolf (1966)
and Sharp (1993)
ultimate shape factor
by
Sharp (1993)
y
= 1.3
1.4
u
u
y
F
F
=
Note: The ultimate shape factor shown here is modified from its
original form to correspond with Equation (3).
Figure 2.3 Limit states for fully compact symmetric cross-sections
The web elements of symmetric cross-sections such as those seen in Figure
2.3a can be simplified into a solid rectangular element under bending as shown in
Figure 2.2a with simply-supported boundary conditions at web-flange junctions. Since
the stress magnitude at the web-flange junction is not significantly different from that
10
at the extreme fiber, the former is assumed to be the same as the latter. This is based
on the assumption that the element thickness is sufficient to prevent buckling.
The shape factor defined in Equation (2) is the ultimate load carrying capacity
divided by the idealized yield capacity (M
y
= F
y
S). Hence, the term shape factor can
also be used for members under other types of loading such as uniform compression.
Examples include the flange elements of sections such as those shown in Figure 2.3a
under bending with respect to the strong axis, assuming buckling is not a problem.
Disregarding the stress variation on flanges, the ratio of the ultimate stress to the yield
stress is the ultimate shape factor for the flange,
u
u
y
F
F
= (4)
while the yield shape factor for the flange is obviously unity.
2.2 Limit State Stresses of Component Elements in the AA Specification
Limit state stress is defined as the estimated actual strength of a component
element based on selection of governing limit state, either yielding, fracture (ultimate),
or buckling, obtained from theoretical and/or experimental studies. As mentioned in
Section 2.1, by simply multiplying the limit state stress with a fixed section property,
either the section modulus or area of the cross-section, the moment capacity or axial
load capacity can be calculated.
Due to uncertainties such as over-loading, construction flaws, and variations of
material properties, in Allowable Stress Design the limit state stress is divided by a
safety factor to obtain the Allowable Stress. In Load and Resistance Factor Design
(LRFD), the limit state stress is multiplied by a resistance factor () to obtain the
Factored Limit State Stress. The resistance factor accounts for the uncertainties
related to strength. Although this study is based on the Allowable Stress Design, the
11
term allowable stress can be replaced with the factored limit state stress for the
Load and Resistance Factor Design.
As summarized in Table 2.2 for the Allowable Stress Design, the AA
Specification (2000a) provides two different limit state stress expressions for tensile
stresses in component elements depending on the limit state that may be yield or
ultimate. The minimum of the two allowable stresses is used in design. In Load and
Resistance Factor Design, the minimum of the two factored limit state stresses is used.
The allowable stresses shown in Table 2.2 incorporate the shape factors for
component elements discussed in Section 2.1. The safety factors on yield and ultimate
strength are denoted by n
y
and n
u
, respectively, while the coefficient k
t
is an additional
safety factor on ultimate strength due to the notch sensitivity of some alloy-temper
combinations, e.g., 2014-T6, 6066-T6, and 6070-T6. Details of the notch sensitivity
are given in the AA Specification (2000a). As seen in Table 2.2, allowable tensile
stress for each component element is constant for a given limit state.
Table 2.2 Options to compute tension allowable stresses of a flexural member by
AA Specification
component element
allowable stress based on
yield limit state
allowable stress based on
ultimate limit state
tension flange of
structural shapes
F
ty
/n
y
F
tu
/k
t
n
u
tension web of
structural shapes
1.30F
ty
/n
y
1.42F
tu
/k
t
n
u
Note: F
ty
= tensile yield stress
F
tu
= tensile ultimate stress
Allowable compressive stress or factored limit state compressive stress varies
according to the equivalent slenderness ratio (
p
), which falls into one of three ranges,
the yielding, inelastic buckling, and post buckling ranges, as shown in Figure 2.4.
12
Figure 2.4 Limit state stress for a component element from the AA Specification
The equivalent slenderness ratio is defined in Equation (5), which is derived from
Equation (6):
( )
2
12 1
, where
p
p
b
t k
| |
= =
|
\ .
(5)
( )
2 2
2 2
2
12(1 ) /
cr p
p
E E
F k
b t
(6)
where k
p
= the plate buckling coefficient depending on the edge conditions and aspect
ratio of the plate, = Poissons ratio, E = Youngs modulus, b = plate width, t = plate
thickness, and F
cr
= buckling stress. For flange and web component elements with
simply-supported boundary conditions at the web-flange junctions, examples of the
equivalent slenderness ratios are shown in Table 2.3.
Table 2.3 Equivalent slenderness ratios in the AA Specification (2000a)
boundary and loading
b
conditions
equivalent
slenderness ratio
plate buckling
coefficient (k
p
)
1.6
p
b
t
| |
=
|
\ .
4.00
5.1
p
b
t
| |
=
|
\ .
a
0.405
0.67
p
b
t
| |
=
|
\ .
23.9
Note: a. 0.405 is back calculated from the equivalent slenderness ratio listed in
Alcoa (1958):
p
= 5.13(b/t)
b. S.S denotes a simply-supported boundary condition.
13
The post-buckling range equation illustrated in Figure 2.4 proposed by
Jombock and Clark (1968) takes into account the non-linear stress distribution of a
buckled plate. The inelastic buckling range equation proposed by Clark and Rolf
(1966) is a linear equation composed of B and D factors, which is a simplification of
the original inelastic plate buckling equation based on experimental studies by Stowell
(1948). Further details are summarized in Kim (2000). The cut-off in the yielding
range is equal to the yield shape factor multiplied by the yield stress, which is based
on the yield limit state. However, experimental studies, such as DOD (1994), show
that the uniaxial stress-strain relationship of aluminum alloys under compression is
very similar to that of tension up to the tension fracture point. Because the uniaxial
stress-strain relationship under compression progresses beyond the equivalent tension
fracture point, it is generally conservative to assume that the ultimate stress and strain
for compression are the same for tension. For these reasons, this study suggests
including the use of a cut-off that is calculated according to the ultimate limit state, in
addition to the yield limit state. In other words, the cut-off should increase to a level
similar to the ultimate limit state of tension elements as shown in Table 2.2. Before
applying the implied tension shape factors in this table to compression, the ultimate
shape factor for web elements will be examined in the following section.
2.3 Rigorous Equation of Ultimate Shape Factor for Rectangular Web Elements
Numerical integration of the non-linear stress distribution when the stresses at
both extreme fibers are at the ultimate stress, as shown in Figure 2.2e, is the most
common approach to obtaining the ultimate shape factor. In this approach, the stress-
strain relationship is approximated by the modified Ramberg-Osgood equation, as
shown in Equation (7):
0.002
n
y
f f
E F
| |
= +
|
|
\ .
(7)
14
where E = Youngs modulus, F
y
= yield stress, = variable strain, f = variable stress
and n = exponent. The original equation is given in Ramberg and Osgood (1943). In
order to use Equation (7), the exponent, n, must be determined. However, although the
AA Specification gives the yield stress, Youngs modulus, and the ultimate stress, the
strain at the ultimate stress (will be referred to as the ultimate strain hereafter,
u
) is
still needed to determine the exponent, n:
( )
log 500 log
u
u
y
F
F
u
E
F
n
| |
|
|
\ .
(
=
(8)
Although statistical data for the ultimate strain is not available, the minimum
percent elongation is available in the Aluminum Standards and Data (AA, 2000c).
Since generally the percent elongation is somewhat larger than the ultimate strain, the
ultimate strain may be estimated from the percent elongation, and hence the exponent,
n, can be determined.
It is a cumbersome task to compute the shape factor using the numerical
integration method due to the iterative nature of the method itself and the high non-
linearity of the modified Ramberg-Osgood equation. For this reason, an analytical
integration is used to derive the closed-form shape factor equation in this study. As
shown by Eberwien and Valtinat (2001), the moment capacity for a rectangular solid
member with a non-linear stress distribution expressed by the modified Ramberg-
Osgood equation can be obtained through the Bernoulli hypothesis.
Figure 2.5 Strain and stress distributions for a rectangular aluminum section
15
With the hypothesis, a linear strain distribution can be assumed, as shown in
Figure 2.5b. Using the stress-strain relationship given in Equation (7),
2 2
0.002
n
u u y
h h f f
y
E F
| |
= = +
| `
|
\ .
)
(9)
and the derivative of Equation (9) is
1
2 2 1
500
n
n
u u y
h h n f
dy d df
E F
= = +
`
)
(10)
The ultimate moment capacity expression is derived by introducing Equations (9) and
(10) into the following integral expression using notations from Figure 2.5. Since a
typical stress-strain relationship from uniaxial tension tests is quite similar to that from
compression tests as shown in DOD (1994), the integral for the upper region is
doubled to obtain the complete ultimate moment capacity:
/ 2
0
2
h
u
M b f y dy =
(11)
In this integral, the h/2 limit is replaced with F
u
due to a change in variable from y to f.
The final closed-form ultimate moment capacity is shown in Equation (12).
2
2
2
2 2
1 1 1 1
2 3 500 2 1 500 2
n n
u u u u u
u
u y y
bh F F F F F n n
M
E n F n F E
(
| | | |
+ | | | | | | | |
(
= + +
| |
| | | |
| |
+ + (
\ . \ . \ . \ .
\ . \ .
(12)
2
6
y
y
bh F
M = (13)
The closed-form ultimate shape factor for a rectangular aluminum solid section
is obtained by normalizing the ultimate moment capacity of Equation (12) using the
yield moment capacity of Equation (13):
2
2
2 2
3 1 1 1 1
3 500 2 1 500 2
n n
u u u u u u
u wo
y u y y y
M F F F F F n n
M F E n F n F E
(
| | | | | |
+ | | | | | | | |
(
= = = + +
| | |
| | | |
| | |
+ + ( \ . \ . \ . \ .
\ . \ . \ .
(14)
16
where
wo
is an alternative denotation for emphasizing that the shape factor is limited
to a symmetric section. For an unsymmetric section, another denotation is used,
w
, as
shown in Equation (50).
As seen in DOD (1994) and uniaxial tension tests performed in this study, the
typical values of the ultimate strain (
u
) for 6061-T6 are between approximately 6 and
8%. For the same material, the typical value of percent elongation is 12% as tabulated
in AA (2000b). Similarly, since the minimum value of the percent elongation is 8%
for 6061-T6 as shown in AA (2000c), it is reasonable to predict that the minimum
value of the ultimate strain be in the range of 4 to 6%. To evaluate the effect of a
variation in the ultimate strain, shape factors are computed based on various ultimate
strain values using Equation (14), with results shown in Figure 2.6. The shape factor
variation is less than 1% for ultimate strain values that are within 50% larger or
smaller than the minimum percent elongation.
Note:
u8%
is
u
when
u
= 8%.
Figure 2.6 Variation of the ultimate shape factor for a rectangular web element
with respect to the ultimate strain (6061-T6)
Thus, the aforementioned assumption that the ultimate strain may be estimated
from the percent elongation is reasonable. In this study, the recommended value for
the ultimate strain is between half and equal to the minimum percent elongation. The
ultimate shape factors are computed using Equation (14) for some 6000 series alloys
17
in Table 2.4, which are more frequently used for building construction than any other
alloys.
Table 2.4 Material properties and ultimate shape factors for solid rectangular
sections for some 6000 series alloys (extrusions)
Alloy-temper
a
F
y
(MPa)
b
E
(MPa)
a
F
u
(MPa)
c
u
d
n
e
u
=
wo
f
1.42
u
tu ty
F F
1.25 0.2
tu
u
ty
F
F
= +
1.42
tu
u
ty
F
F
=
u
Equation (14)
tu ty
F F
20
elements with 4 integration points are used to take into account a large variation of
thickness. The alloy-temper is assumed to be extruded 6061-T6 with the minimum
material properties given in Table 2.4 except for the ultimate strain. The two
recommended ultimate strain values, 4% and 8%, which are half and equal to the
percent elongation, respectively, are used to observe their effect on the limit state
stress. As previously stated, the ultimate strain variation affects the Ramberg-Osgood
exponent (n), which in turn affects the ultimate shape factor for the web (
u
).
Since this study deals with monotonic loading, the isotropic hardening model
is used. In this case, once a stress reaches the ultimate stress, the stress remains
constant as the plastic strain exceeds the ultimate strain. Due to uncertainty after this
stage, it is assumed that a whole member reaches failure when the Von Mises stress at
a single point of the member reaches the ultimate stress. This occurs when a member
is too compact to buckle.
On the other hand, the failure of a member can also be initiated by buckling
when the member is less compact. In this case, the peak load of the member is
obtained before any point of the member reaches the ultimate stress. In this study,
these two possibilities of failure are considered simultaneously so as to find an
ultimate load factor.
The boundary conditions for component elements of doubly symmetric I-
shaped sections are idealized as shown in Figure 2.8. To avoid singularity of the
stiffness matrix, one longitudinal degree of freedom is restrained at the span center.
Equal and opposite loadings are applied at the loaded edges, where rigid beam
elements are attached. Elastic eigen-value analyses to generate initial geometric
imperfections precede the non-linear analyses. The maximum imperfection amplitude
is determined based on the standard flatness tolerance provided by AA (2000c).
21
Figure 2.8 Boundary and loading conditions for finite element analyses
Figure 2.9 Parametric study results for component elements (a) web (b) flange
From the results of the parametric study for component elements shown in
Figure 2.9, the following modifications of the limit state stress equations are made for
the ultimate limit state. For a very low width-to-thickness range, the yield stress times
the ultimate shape factor is employed for the raised cut-off. For the transition range
between the raised and current cut-off values, the equation line of the inelastic
buckling range (S
1
width/thickness S
2
) is extended to the raised cut-off value with
the same slope.
For the material used in this parametric study, the ultimate shape factor is
1.598 for the web element using Equation (14) (
u
= 4%) and 1.086 for the flange
22
element using Equation (4). These modifications work well with both series of finite
element analyses based on the two ultimate strain values, 4% and 8%. This parametric
study confirms that the ultimate limit state can be considered for compressive
component elements. The results of Kim (2000) are reevaluated in the present work
and the equations of Kim (2000) are modified to what is shown in Table 2.5.
Table 2.5 (a) Limit state stress equations and limits and (b) shape factors
(a)
limit state stress
1
b
S
t
limit
S
1
limit state stress
1 2
b
S S
t
limit
S
2
limit state stress
2
b
S
t
F
p
=
F
cy
cy
B F
D
p
b
F B D
t
=
1
k B
D
2
p
k BE
F
b
t
=
(b)
component element yield limit state ( =
y
) ultimate limit state ( =
u
)
Flange 1.0
tu cy
F F
Web 1.3
1.25 0.2
tu cy
F F +
Note. a. F
p
= flange limit state stress (F
f
) or web limit state stress (F
w
).
b. B,D, k
1
and k
2
are factors provided in the AA Specification (2000a). The factors for a
flange differ from those for a web.
c. See Equation (5) for (Table 2.3).
d.
u
= Equation (4) for flange and Equation (16) for web
e. F
cy
= compressive yield stress
2.6 Moment Capacity Evaluation Approaches
2.6.1 Minimum Moment Capacity Approach (MMCA)
In AA (2000b), to compute the allowable moment capacity of a member, the
allowable stress (or factored limit state stress) of each component element is first
computed. Then from the calculated allowable stresses (or factored limit state
stresses), the minimum is selected and multiplied by the elastic section modulus of the
entire cross-section (S). For the section given in Figure 2.10,
23
M
u
= min(F
f
, F
w
) S (17)
where F
f
and F
w
= allowable stress (or factored limit state stress) for the flange and
web, respectively. However, instead of the use of the allowable stresses (or factored
limit state stresses) for F
f
and F
w
, limit state stress values have been employed in the
parametric study that follows in Section 2.7. This approach is demonstrated by AA
(2000b) through the Illustrative Examples of Design (Part VIII, Example 21) for
symmetric sections with respect to the bending axis.
This approach is denoted by the Minimum Moment Capacity Approach or
MMCA hereafter. Since possible interactions and stress redistributions between
component elements are disregarded, this approach is expected to be rather
conservative.
2.6.2 Weighted Average Stress Approach (WASA and WASA2)
Another method is given in the AA Specification (2000a), where the limit state
stresses obtained from all component elements in each of the compression and tension
sides can be averaged according to contributory area. The averaged stress is multiplied
by the section modulus to compute the moment capacity. This is called the Weighted
Average Stress Approach or WASA in this study. The weighted average stress
equation was first introduced by Jombock and Clark (1968) to compute the crippling
strength of aluminum trapezoidal formed sheet members:
1
3
1
3
f f w w
WT
f w
F A F A
F
A A
+
=
+
(18)
where A
f
and A
w
= flange and web area for either compression or tension (indicated in
Figure 2.10). Sooi and Pekz (1993) extended the WASA to sections with edge-
stiffened and intermediately-stiffened elements by adding further terms for stiffeners
to Equation (18). See Table 4.3 for details. Although the WASA was verified through
experiments in the aforementioned studies, a theoretical basis for the approach has
24
never been investigated. Thus, the accuracy of this method is questionable for
geometric shapes other than tested.
Figure 2.10 Contributions of element groups to the entire moment capacity
The theoretical basis of the WASA is investigated in this study for a doubly
symmetric I-section shown in Figure 2.10. Multiplication of both the denominator and
numerator of Equation (18) by h
c
2
/2 and subsequent simplification of this equation
through an assumption that the flange thickness is relatively small result in Equation
(19):
2 2
f w
c u c
WT
f w
M M
h M h
F
I I I
+
| | | |
=
| |
+
\ . \ .
(19)
where M
f
, M
w
,
M
u
= moment capacity of the (top and bottom) flanges, web, and entire
cross-section respectively, I
f
, I
w
, I = moment of inertia with respect to the neutral axis
of the entire cross-section of the (top and bottom) flanges, web, and entire cross-
section, respectively. For the section in Figure 2.10, I
w
= t
w
h
o
3
/12 and I
f
= I I
w
.
Equation (19) implies that the weighted average stress is an approximately linearized
bending stress measured at the mid-thickness of the flange. Therefore, to obtain an
accurate total moment capacity, the correction shown in Table 2.6 should be made:
25
Table 2.6 Correction in the current WASA
current WASA
(WASA)
proposed WASA
(WASA2)
M
u
= (F
WT
)(S)
M
u
= (F
WT
)
c
h
h
| |
|
\ .
(S)
Note: S = section modulus = I
/(h/2)
The correction in Table 2.6 is insignificant when the flange thickness is
relatively small, such as those of the thin-walled cold-formed sheet members used by
Jombock and Clark (1968) and Sooi and Pekz (1993). Sections consisting of
component elements with large width-to-thickness ratios are not suitable for extrusion
due to production limits (for details, see Kissell and Ferry, 1995). For this reason,
most standard sections listed in AA (2000b) are made of relatively thick component
elements, falling into the yielding or, at least, inelastic buckling ranges. Thus, the
modifications shown in Table 2.5 as well as Table 2.6 are significant for common
aluminum extrusions.
2.6.3 Total Moment Capacity Approach (TMCA)
As an alternative to the WASA2, the Total Moment Capacity Approach
(TMCA) is also proposed in this study. In this approach, the actual non-linear stress
distribution is artificially divided into two linear stress distributions, as shown in
Figure 2.10. Such approximation is consistent with the current AA Specification
(2000a), which is implicit in the WASA. In this figure, the web area and the flange
area are assigned separate linear stress distributions so that no overlap can occur. For
more complicated cross-sections, the cross-sectional area is divided into a web group
and a flange group (for details, see Section 4.4.3). The section modulus of each group
is computed with respect to the neutral axis of the entire cross-section, which is at
mid-depth for the example I-section. The limit state stress computed for each group is
multiplied by the corresponding section modulus to compute the moment capacity.
26
Afterwards, the moment capacities from all groups are added to obtain a member
moment capacity.
For example, the moment capacity of the example I-shaped section can be
expressed as Equation (20) based on the TMCA.
2 2
f f
w w
u f w f f w w
c o
F I
F I
M M M F S F S
h h
= + = + = + (20)
where S
f
, S
w
= the elastic section modulus with respect to the neutral axis of the entire
cross-section of the flange and web groups, respectively. For purposes of comparison
with the weighted average stress equation of Equation (18), another weighted average
stress equation can be defined using the TMCA:
f f w w
TM
F S F S
F
S
+
= (21)
where F
TM
denotes the weighted average stress based on the contributory section
moduli according to the TMCA. Since all of the component elements contribute to the
moment capacity in both the WASA2 and TMCA, these two methods are expected to
be more accurate than the MMCA. In addition, the expression for the TMCA
resembles that for the WASA, which implies that the WASA is a special case of the
TMCA.
Table 2.7 Member allowable stress (or factored member limit state stress) for
simple sections
approach member allowable (or factored member limit state) stress moment capacity
MMCA ( )
min ,
MIN f w
F F F =
M = F
MIN
S
WASA
1
3
1
3
f f w w
WT
f w
F A F A
F
A A
+
=
+
M = F
WT
S
WASA2
1
3
2
1
3
f f w w
WT
f w c
F A F A
h
F
A A h
+ | |
=
|
+
\ .
M = F
WT2
S
TMCA
f f w w
TM
F S F S
F
S
+
= M = F
TM
S
27
The member allowable (or factored member limit state) stress is defined as the
combination of each component element allowable (or factored limit state) stress, and
is expressed based on the approaches described in this section, which is summarized
in Table 2.7. Based on the selected member allowable stress (or factored member limit
state stress) expression, the moment capacities are computed for both the compression
and tension sides. The minimum of the two moment capacities of the two sides is
determined as the moment capacity of the member. When compressive component
elements are in the yielding range, the WASA2 and TMCA can result in both of the
extreme fiber flange stresses being larger than the maximum limit state stress, for the
limit state stress is assumed to occur at mid-thickness of the flange and not the
extreme fiber. Though this assumption does not concede to the initial assumptions
made concerning the yield and ultimate limit states shown in Figure 2.3, this should
not be a problem for simple symmetric sections such as those in Figure 2.1. This is
because the actual non-linear stress distribution on the flange is almost uniform as
seen in Figure 2.10.
Instead of the use of the allowable stresses (or factored limit state stresses) in
Table 2.7, limit state stress values have been employed in the parametric study that
follows in Section 2.7.
2.7 Parametric Study of I-Shaped Sections
To validate the improvements made in Sections 2.3 to 2.6, a parametric study
is conducted for doubly symmetric I-shaped sections, which includes the results of the
finite element analyses conducted in Kim (2000). The width of the entire flange (w
in Figure 2.10) and the depth between the center-lines of flanges (h
c
in Figure 2.10)
are both held constant at 254 mm (10 in.), while the uniform component element
thicknesses are varied to allow a wide range of width-to-thickness ratios, so that most
of the standard sections listed in AA (2000b) can be covered. The length of the
28
members is fixed at 2540 mm (100 in.). The boundary conditions are determined as
shown in Figure 2.11. Other details regarding the finite element model are the same as
for the parametric study for component elements performed in Section 2.5.
Figure 2.11 Model geometry of an I-shaped section for parametric study
The analysis is comprised of three steps. The first step investigates the effect
of the use of the ultimate limit state rather than the yield limit state that is used in the
AA Specification (2000a), as shown in Figure 2.12. The approach abbreviations used
in Figure 2.12 to Figure 2.14 are as follows. The first two categories, AA-Y and AA-
U, refer to the AA Specification equations based on either the Yield or Ultimate limit
state specified in the yielding range (b/t S
1
) as shown in Table 2.5. The two limit
states are distinguished by whether or not the raised cut-off and the extended inelastic
buckling range equations are used. The remaining category indicates the moment
capacity evaluation approach selected, as defined in Section 2.6.
The horizontal axis of Figure 2.12 to Figure 2.14 is the slenderness factor,
y cr
F F = (22)
which indicates a general sense of the slenderness of the component plate elements of
a cross-section. The slenderness factor is not used in design procedures, but for visual
convenience. F
cr
denotes the minimum local buckling stress from numerical buckling
analysis for the entire cross-section. The vertical axis is the moment capacity obtained
29
from the finite element analysis divided by the moment capacity obtained with each
approach specified in the graph. As seen in this figure, an approximately 7%
difference in the member capacity is observed due to consideration of the ultimate
limit state. The data variation also decreases when the ultimate limit state is
considered.
0.3 0.4 0.5 0.6 0.7 0.8
0.8
0.9
1
1.1
1.2
1.3
M
F
E
M
/
M
a
p
p
r
o
a
c
h
e
s
M
FEM
/ M
approaches
mean c.o.v
M
FEM
/ M
AA-Y-WASA
M
FEM
/ M
AA-U-WASA
1.209 0.056
1.141 0.029
=
y cr
F F
Figure 2.12 Effect of the use of the ultimate limit state (
u
= 4%)
0.3 0.4 0.5 0.6 0.7 0.8
0.8
0.9
1
1.1
1.2
1.3
M
F
E
M
/
M
a
p
p
r
o
a
c
h
e
s
M
FEM
/ M
approaches
mean c.o.v
M
FEM
/ M
AA-U-WASA
M
FEM
/ M
AA-U-WASA2
1.141 0.029
1.04 0.024
=
y cr
F F
Figure 2.13 Effect of the modification of WASA (
u
= 4%)
The second step investigates the effect of the modification of WASA in the
AA Specification (2000a), as shown in Figure 2.13. The difference from the finite
30
element analysis is significantly decreased by approximately 10% due to the
modification in WASA.
The third and final step compares the two currently available approaches (M
AA-
Y-MMCA
and M
AA-Y-WASA
) in the AA Specification (2000a) and those developed in this
study (M
AA-U-WASA2
and M
AA-U-TMCA
). As seen in Figure 2.14, the approaches
developed in this study predict the ultimate-plastic capacity more precisely than the
current approaches in the AA Specification (2000a) for varied slenderness.
Since the difference between the WASA2 and TMCA is insignificant, either
approach can be employed for the type of cross-sections considered in this chapter.
The moment capacities obtained from all four approaches, as well as from the finite
element analyses are listed in Table 2.8. All of the finite element computations shown
in Figure 2.12 to Figure 2.14 are made at an ultimate strain (
u
) of 4%. The
percentages indicated in the subscripts of the FEM analyses in Table 2.8, 4% and 8%,
represent the ultimate strain values used. When the ultimate strain is 8%, results do
not change significantly from that of 4% as seen in this table. This is consistent with
the conclusion made from Figure 2.6.
0.3 0.4 0.5 0.6 0.7 0.8
0
0.5
1
1.5
2
M
F
E
M
/
M
a
p
p
r
o
a
c
h
e
s
M
FEM
/ M
approaches
mean c.o.v
M
FEM
/ M
AA-Y-MMCA
M
FEM
/ M
AA-Y-WASA
M
FEM
/ M
AA-U-WASA2
M
FEM
/ M
AA-U-TMCA
1.248 0.055
1.209 0.056
1.04 0.024
1.037 0.023
M
TEST
/ M
approaches
Figure 2.14 Comparison between the current and proposed approaches (
u
= 4%)
=
y cr
F F
31
Table 2.8 Parametric study results
t
w
(mm)
t
f
(mm)
AA-Y-MMCA
y
M
M
AA-Y-WASA
y
M
M
AA-U-WASA2
y
M
M
AA-U-TMCA
y
M
M
FEM-4%
y
M
M
FEM-8%
y
M
M
y
cr
F
=
F
25.4 42.3 1.000 1.023 1.309 1.316 1.375 1.378 0.233
25.4 31.8 1.000 1.031 1.277 1.285 1.347 1.340 0.289
25.4 21.2 1.000 1.047 1.223 1.231 1.290 1.280 0.384
25.4 15.9 0.971 1.037 1.157 1.164 1.240 1.230 0.473
25.4 12.7 0.892 0.990 1.105 1.111 1.200 1.190 0.566
12.7 42.3 1.000 1.012 1.289 1.292 1.314 1.305 0.259
12.7 31.8 1.000 1.017 1.251 1.255 1.280 1.270 0.335
12.7 21.2 1.000 1.025 1.171 1.175 1.210 1.200 0.469
12.7 15.9 0.954 0.992 1.085 1.088 1.150 1.140 0.581
12.7 12.7 0.871 0.929 1.013 1.015 1.090 1.090 0.679
8.46 42.3 1.000 1.008 1.282 1.284 1.290 1.275 0.268
8.46 31.8 1.000 1.011 1.241 1.244 1.256 1.240 0.351
8.46 21.2 1.000 1.017 1.151 1.154 1.180 1.170 0.506
8.46 15.9 0.948 0.975 1.056 1.058 1.100 1.100 0.644
8.46 12.7 0.863 0.905 0.975 0.976 1.020 1.030 0.767
6.35 42.3 1.000 1.006 1.276 1.278 1.276 1.260 0.349
6.35 31.8 1.000 1.009 1.233 1.236 1.240 1.220 0.358
6.35 21.2 1.000 1.013 1.136 1.138 1.170 1.150 0.524
6.35 15.9 0.945 0.966 1.034 1.036 1.080 1.080 0.677
6.35 12.7 0.860 0.892 0.946 0.947 0.998 1.000 0.817
2.8 Experiments and FEM Simulation
To support the approaches developed in this study, physical tests have been
conducted for three doubly symmetric Aluminum Association Standard I-Beams in
AA (2000b): I-3x1.64. The alloy-temper of the specimens is 6063-T6, of which the
minimum material properties are listed in Table 2.4. Since the study is based on the
strength of component elements, continuous lateral supports are required. However,
such supports are practically impossible to construct in physical tests.
Note: All dimensions are in mm and not to scale
Figure 2.15 (a) Dimensions of section -3x1.64 (b) schematic test setup
32
(a)
(b)
(c) (d)
Note: all dimensions are in mm and not to scale
Figure 2.16 Test setup (a) plan view (b) longitudinal section (c) cross-section (d)
Detail A
33
For this reason, a parametric study has been conducted using the finite element
method to find appropriate lateral support spacing so that the ultimate load factor and
the corresponding displacement can be similar to those of the continuous lateral
support case. The study results show 304.8 mm (12 in.) is adequate for the lateral
support spacing. The test setup is determined as shown in Figure 2.15, including
dimensions for the tested specimens. Details of the test setup are shown in Figure
2.16.
All finite element modeling issues covered are similar to those for the
parametric study in Section 2.7 except for the following. First, applied load is not pure
bending but two-point bending as shown in Figure 2.15b. Second, the lateral support
spacing is not continuous. Third, bi-linear spring elements are attached between the
spreader plates and the specimen, so that only compression can be transferred. This is
to simulate the contact behavior of the actual test setup, in which the spreaders were
simply placed on the specimen without any moment connections such as welding or
bolting. Fourth, the median of five uniaxial tension test results (Table 6.27a in the
Appendix) obtained from one of the specimens is introduced into the finite element
analyses. Fifth, different finite element models are used. The model using four-noded
linear shell elements with reduced integration is denoted by SHELL; and the model
using twenty-noded quadratic hexahedral solid elements with reduced integration is
denoted by SOLID. The SOLID model uses two layers of solid elements in the
thickness direction. Sixth, initial geometric imperfections are generated using elastic
eigen-value analyses with a maximum amplitude based on either the maximum value
from the actual measurements in this study (0.048 mm, model FEM 1) or the standard
flatness tolerance (0.127 mm, model FEM 2 to 3) according to AA (2000c). The actual
measurements for geometric imperfections are shown in Figure 6.6 in the Appendix.
34
When specimens are removed from the test frame, residual deformation
remains, as shown in Figure 2.17. A single wave is formed near the span center in
each specimen. The deformed shape near failure from one of the finite element
simulations, as shown in Figure 2.18, is similar to those from the physical tests in
Figure 2.17.
(a)
(b)
Figure 2.17 Residual deformation of tested specimens (I-3x1.64) (a) side (b) plan
Figure 2.18 Deformed shape near failure using finite element method (SOLID)
35
As plotted in Figure 2.19, the load factor-displacement curves obtained from
both the physical tests and the finite element simulations show close agreement with
each other. The variation in the test results are most likely due to the variations in the
material properties, as shown in Table 6.27a and Table 6.27b in the Appendix. The
average of the maximum load factors from the physical tests is compared to the
current specification approaches and those developed in this study within the dashed-
line oval of Figure 2.14. The test results follow the trend of the parametric study,
which further validates the two proposed approaches in this study. The proposed
approaches are based on the ultimate limit state using the WASA2 or TMCA to
integrate the moment capacities from individual component elements.
0 50 100 150
0
0.2
0.4
0.6
0.8
1
1.2
1.4
DESCRIPTION IMP MAX-LF DISP-XL MODEL
TEST 1
TEST 2
TEST 3
FEM 1
FEM 2
FEM 3
0.048
0.127
0.127
1.2
1.275
1.301
1.238
1.218
1.204
109.2
120.7
127.7
121.2
101.9
131.1
SHELL
SHELL
SOLID
DISP-XL = Displacement at the maximum LF
SCVD (Span Center Vertical Displacement, mm)
L
F
=
M
u
/
M
y
Figure 2.19 Comparison of load factor-displacement results for -3x1.64 sections
2.9 Application to the AA Specification
This study shows that the proposed approaches based on the ultimate limit
state are in good agreement with both the finite element simulations and physical tests.
However, it is desirable to maintain a certain factor of safety on yield strength in
actual designs. As summarized in Table 2.9, the AA Specification (2000a) requires
choosing the minimum for the allowable stresses (or factored limit state stresses)
based on the yield and ultimate limit states for each tension component element.
36
However, there is only one allowable stress (or factored limit state stress), which is
based on the yield limit state, for compression component elements.
Procedure I, which is one of the proposed approaches shown in Table 2.9, is
almost the same as the current AA Specification, except that allowable stresses (or
factored limit state stresses) for both limit states are given not only for the tension but
also for the compression component elements as a result of the development in this
study. The allowable stress (or factored limit state stress) based on the yield limit state
is referred to as the yield allowable stress (or yield factored limit state stress) (F
ay
),
and the allowable stress (or factored limit state stress) based on the ultimate limit state
is referred to as the ultimate allowable stress (or ultimate factored limit state stress)
(F
au
) hereafter. In this approach, the same factor of safety on yield strength (n
y
) is
maintained as in the AA Specification (2000a), with a value of 1.65 for building and
similar type structures.
Although Table 2.9 and Table 2.10 are based on the Allowable Stress Design,
these tables can also be applied to the Load and Resistance Factor Design by replacing
the reciprocal of each safety factor with the corresponding resistance factor.
However, it does not seem reasonable to employ a uniform safety factor on
yield strength regardless of the margin between the yield and ultimate stresses for a
wide variety of alloy-temper combinations. As seen in Figure 2.7, the ratio of the
ultimate to yield stress varies from 1.07 to 3.67. The materials with larger margins
should be safer than those with smaller margins, e.g., see Figure 1.1. Thus, an
alternative approach to computing the allowable stress is proposed in Table 2.9, and is
named Procedure II. In this approach, 25% of the margin between the yield and
ultimate allowable stresses is added to the yield allowable stress. Thus, the allowable
stress increases as the ratio of the ultimate to yield stress increases, which results in a
varying safety factor on yield strength.
37
Table 2.9 Governing allowable stresses
approaches
governing allowable stress for
tension
governing allowable stress for
compression
current AA
Specification
( )
min ,
a ay au
F F F =
F
a
= F
ay
Procedure I
( )
min ,
a ay au
F F F = (23)
proposed
approaches Procedure II
( ) ( )
0.25 min 1.25 ,
a ay au ay ay au
F F F F F F = + (24)
Note: F
a
= the governing allowable stress
F
ay
= the member allowable stress based on the yield limit state. See Table 2.7.
F
au
= the member allowable stress based on the ultimate limit state. See Table 2.7.
Table 2.10 Allowable stress equations for (a) tension component element,
(b) compression component element, and (c) shape factors
(a)
AA Section allowable stresses
ay y ty y
F F n =
3.4.2
3.4.4
( )
au u ty t u
F F k n =
(b)
AA
Section
allowable stresses in
yielding range
b/t S
1
limit
S
1
allowable stresses in
inelastic buckling range
S
1
b/t S
2
limit
S
2
allowable stresses in
post buckling range
S
2
b/t
y cy
ay
y
F
F
n
=
y cy
B F
D
3.4.15
3.4.16
3.4.18 u cy
au
u
F
F
n
=
y
u
n
u cy
n
B F
D
1
ay au
y
F F
b
B D
n t
= =
| |
|
\ .
1
k B
D
2
ay au
y
k BE
F F
b
n
t
= =
(c)
AA Section yield shape factor ultimate shape factor
3.4.2
1.0
y
=
u tu ty
F F =
3.4.4
1.3
y
= 1.25 0.2
u tu ty
F F = +
a
3.4.15, 3.4.16
1.0
y
=
u tu cy
F F =
b
3.4.18
1.3
y
= 1.25 0.2
u tu cy
F F = +
c
Note: a, c. This is for a symmetric section. See Table 4.8c for an unsymmetric section.
a. In the AA Specification,
u
= 1.42F
tu
/F
ty
.
b, c. Not available in the AA Specification.
For other coefficients, see the AA Specification. Each allowable stress in this table is
introduced into equations in Table 2.7 to compute the member allowable stress.
38
The safety factor on ultimate strength (n
u
= 1.95 for building and similar type
structures) given in the AA Specification (2000a) is maintained, since only the
specified percentage of the ultimate allowable stress is used. Since the safety factor on
yield strength is defined as the ratio of the yield limit state stress to the allowable
stress, the following expression characterizes the varying safety factor on yield
strength for a tension side component element:
y y y a
n F F = (25)
1 1.5 2 2.5 3 3.5 4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
2.2
mean stdev min max
n
y
~
1.54 0.129 1.32 2.01
n
y
= 1.65
~
n
y
= 1.32
~
n
y
= 1.00
~
k
t
= 1.00
k
t
= 1.10
k
t
= 1.25
0 200 400
0.6
0.8
1
1.2
1.4
1.6
1.8
2
2.2
(a) (b)
Figure 2.20 Safety factor on yield strength of the tensile allowable stress
(AA Section 3.4.4) for a plate under bending based on Procedure II
In Figure 2.20a, using Equation (25) based on Procedure II, the varying safety
factors on yield strength of a rectangular web element under bending are plotted with
solid circles for the 986 alloy-temper combinations selected in Section 2.4. It is seen
that the average of the varying safety factor on yield strength (1.54) is only 6.7%
smaller than the current fixed safety factor on yield strength (1.65). The solid and
dashed curves represent the analytic expressions of the varying safety factors for the
materials with different notch sensitivities (k
t
). Note that the majority of the alloy-
temper combinations are concentrated near the average, as shown in Figure 2.20b. In
y
n
tu ty
F F numbers of alloy-temper
39
addition, the minimum of the varying safety factor on yield strength is set to 1.32 due
to the first safeguard in Equation (24). The second safeguard in this equation is for the
case that the yield allowable stress is larger than the ultimate allowable stress.
Similarly, varying safety factors on yield strength are computed for I-shaped
sections used in Section 2.7, as seen in Figure 2.21. The varying safety factor on yield
strength for an entire cross-section is defined as follows:
moment capacity based on yield limit state stresses and TMCA
moment capacity based on allowable stresses and given approaches
y
n = (26)
This definition of the varying safety factor for a cross-section composed of web and
flange elements is consistent with that for a rectangular web element in Equation (25).
The only difference is that the moment capacity evaluation approaches in Section 2.6
are incorporated to combine limit state stresses for all component elements. The
numerator is the moment capacity computed by the TMCA when each component
element reaches the yield limit state stress implied in Table 2.2 for the tension side
and defined in Table 2.5 for the compression side. The moment capacity in the
denominator of Equation (26) is calculated as follows; the member allowable stress is
calculated from Table 2.7, which is then used to calculate the governing allowable
stress in Table 2.9, and this is then multiplied by the section modulus. The
abbreviations used in Figure 2.21 are based on the approach combinations used in the
denominator. SF
AA
, SF
P1
and SF
P2
are the varying safety factor on yield strength
computed when the allowable moment capacity is based on the current AA
Specification (2000a) and Procedures I and II in Table 2.9, respectively. The last
category in these abbreviations represents one of the moment capacity evaluation
approaches in Table 2.7. In Figure 2.21, WASA2 is not shown because it is almost
identical to TMCA.
40
0.3 0.4 0.5 0.6 0.7 0.8
1.2
1.4
1.6
1.8
2
2.2
S
a
f
e
t
y
F
a
c
t
o
r
A
g
a
i
n
s
t
Y
i
e
l
d
i
n
g
(
n
y
)
SF
AA-MMCA
SF
AA-WASA
SF
P1-TMCA
SF
P2-TMCA
SF
approaches
mean c.o.v
alloy-temper = 6005-T5
1.974 0.055
1.903 0.069
1.741 0.033
1.741 0.033
(a)
0.3 0.4 0.5 0.6 0.7 0.8
1.1
1.2
1.3
1.4
1.5
1.6
1.7
1.8
1.9
2
S
a
f
e
t
y
F
a
c
t
o
r
A
g
a
i
n
s
t
Y
i
e
l
d
i
n
g
(
n
y
)
SF
AA-MMCA
SF
AA-WASA
SF
P1-TMCA
SF
P2-TMCA
SF
approaches
mean c.o.v
alloy-temper = 6063-T5-b
1.852 0.031
1.804 0.04
1.65 0
1.615 0.012
(b)
Figure 2.21 Comparison of safety factor on yield strength between current and
proposed approaches for I-sections of Section 2.7
Two types of materials are used in Figure 2.21. In Figure 2.21a, the first group
(6005-T5, 6105-T5, 6351-T5, 6061-T6, 6061-T6510, and 6061-T6511) from among
the 6000 series alloy-temper combinations in Table 2.4 is used. When the width-to-
thickness ratio (b/t) of a compression flange falls into the yielding range as specified
alloy-temper
6005-T5, 6105-T5, 6351-T5
& 6061-T6, -T6510, -T6511
=
y cr
F F
alloy-temper
6063-T5
=
y cr
F F
41
in Table 2.10b, the ultimate limit state will govern for this material, since the ultimate
to yield stress ratio of this material group (F
tu
/F
cy
= 1.086) is smaller than the ratio of
the safety factor on ultimate strength to the safety factor on yield strength (n
u
/n
y
= 1.18
for building type structures). Further details are as follows: a compression flange has
two allowable stresses as shown in the yielding range of Table 2.10b, F
ay
= F
cy
/n
y
and
F
au
= F
tu
/n
u
. If the ultimate limit state governs, then F
au
F
ay
. Thus, F
tu
/n
u
F
cy
/n
y
,
and then, F
tu
/F
cy
n
u
/n
y
. In this computation, all stresses are positive. However, if the
width-to-thickness ratio (b/t) falls outside of the yielding range, the yield limit state
may govern, since the two allowable stresses for a compression flange become closer
as shown in the inelastic buckling range of Table 2.10b. For this reason, regardless of
the slenderness factor, the varying safety factor on yield strength is larger than or
equal to 1.65, as in the AA Specification. So for this group there is no difference
between Procedures I and II.
In Figure 2.21b the third alloy-temper combination (6063-T5) in Table 2.4 is
used. Similarly, since the ultimate to yield stress ratio (F
tu
/F
cy
= 1.4) is larger than the
ratio of the safety factors (n
u
/n
y
= 1.18), the yield limit state governs for a compression
flange. Thus, regardless of the slenderness factor, the safety factor on yield strength is
a constant, 1.65 if Procedure I is used. When Procedure II is used, the safety factor
varies, as expected.
In the aforementioned discussion, the governing limit state should also be
examined with respect to a compressive web component element. However, with these
types of cross-sections and material properties, it is found that the effect of the web
element is insignificant, as shown in computational examples provided in the
Appendix.
In Figure 2.21, it is found that the calculated member moment capacity by
Procedure I is 9 to 13% more than that can be computed by the current AA
42
Specification. A constant safety factor on yield strength of 1.65 is maintained.
Procedure II further increases the calculated member moment capacity for materials
with a larger margin between the ultimate and yield stresses. Although this will result
in a variation in the safety factor on yield strength, it should not cause any significant
problem for maintaining a desirable strength margin for safety. Step-by-step
computation procedures for the allowable moment capacity and the safety factor are
provided in the Appendix for a sample cross-section.
2.10 Conclusions
For tension component elements, the AA Specification provides two different
allowable stress (or factored limit state stress) expressions at the yield and ultimate
limit states. However, for compression component elements, the allowable stress (or
factored limit state stress) is provided only for the yield limit state. In this study, the
limit state stress at the ultimate limit state is developed for compression component
elements to take advantage of the ultimate-plastic capacity of compact aluminum
members in bending.
An expression for the ultimate limit state stress for compression component
elements is obtained from rigorous closed-form expressions derived for the ultimate
shape factor for rectangular web elements integrating the stresses across the cross-
section. The rigorous expression thus obtained is simplified for practical design
purposes. A parametric study for component elements is conducted using the finite
element method. Based on the above studies, the ultimate limit state stress equations
are proposed for compression component elements. In addition, the empirically
developed weighted average stress approach (WASA) is investigated, and a simple
modification is proposed. Moreover, a more general approach denoted by the Total
Moment Capacity Approach (TMCA) is developed.
43
The proposed approaches are more accurate for calculating the moment
capacities of laterally supported flexural members. This is validated through a
parametric study of I-shaped sections with a wide range of slenderness, using the
finite element method. The proposed approaches are further supported by the flexural
tests of standard sections.
For design purposes, two procedures are suggested. In Procedure I, the factor
of safety on yield strength used in the current specification is maintained. In Procedure
II, the safety factor is applied to the additional inelastic capacity. Procedure II is more
reasonable in that the safety factor varies depending on the margin between the
ultimate and yield stresses.
44
3. SYMMETRIC CROSS-SECTIONS WITH TAPERED THICKNESS
COMPONENT ELEMENTS
Some standard sections listed in AA (2000b) have tapered thickness plate
flanges, as shown in Figure 3.1a. Also custom shapes with tapered thickness flanges
may be extruded as shown in Figure 3.1b.
(a) (b)
Figure 3.1 Examples of sections with tapered elements (a) standard extruded
shapes (b) custom extruded shapes
The behavior of a component element with tapered thickness has been investigated for
aircraft and structural member analysis by Pines et al. (1947), Wittrick et al. (1962),
Chehil et al. (1973), Kobayashi et al. (1990), Mizusawa (1993) and Ohga et al. (1995).
However, The behavior of tapered thickness elements has not been fully incorporated
in both AISC (1998) and the AA Specification (2000a). Instead, an average thickness
is used in design.
In this study, stiffness matrices for plates with tapered thickness are derived for
use in the framework of an available finite strip analysis program. Using the program,
plate buckling coefficients can be calculated taking into account thickness variation
and boundary conditions. The plate buckling coefficients obtained are introduced into
the current specification approach for aluminum structures to calculate more precise
limit state stresses.
44
45
3.1 Introduction
Sections with tapered plate thickness shown in Figure 3.1a are commonly used
sections made of steel or aluminum. However, the design equations for the behavior of
such tapered plate elements have not been fully developed in specifications. In both
AISC (1998) and the AA Specification (2000a), the treatment of a tapered component
element as a uniform one using an average thickness is implied.
The buckling behavior of a plate element with tapered thickness has been
studied primarily for aircraft design by Pines et al. (1947), Wittrick et al. (1962), and
Chehil et al. (1973). The main focus has been on the buckling of a plate with a
thickness variation in the same direction as the loading. However, since a plate
element that is a component of a structural member is tapered in the direction
perpendicular to loading as shown in Figure 3.2, these studies do not apply to the type
of problem covered in this study.
Figure 3.2 A tapered element separated from an I-section under bending
More recently, Kobayashi et al. (1990), Mizusawa (1993) and Ohga et al.
(1995) have studied the behavior of plate elements tapered in the direction
perpendicular to loading which is the same case as the problem in Figure 3.2.
46
However, the studies have focused on demonstrating theoretical approaches without
any application to structural design codes.
In this study, stiffness matrices are derived for a plate element with tapered
thickness based on the finite strip method. The derived stiffness matrices are
incorporated into the CUFSM program developed by Schafer (1997) to compute the
elastic buckling stress of arbitrary shapes containing tapered thickness component
elements such as those in Figure 3.1. The program is also used to compute the plate
buckling coefficients of an individual tapered thickness plate element with idealized
boundary conditions. The plate buckling coefficients determined in this way are
incorporated in the AA Specification (2000a) so that the limit state stresses of tapered
plate elements are obtained rigorously instead of treating such elements as having a
uniform thickness.
A parametric study of a wide variety of I-shaped sections with tapered
thickness is conducted using non-linear inelastic finite element analyses to validate the
limit state stress expressions based on the rigorously obtained plate buckling
coefficients. Physical tests further support this approach. Allowable stress design
equations are also suggested for application to the AA Specification (2000a) based on
the proposed limit state stress equations. The framework for uniform thickness
sections is maintained in the proposed approach for tapered thickness sections.
3.2 Stiffness Matrices for a Tapered Finite Strip Element
The CUFSM program computes the elastic buckling load factor and
corresponding mode shapes using the finite strip method. However, since CUSFM is
based on the uniform plate theory, a thickness variation cannot be considered directly.
This study derives stiffness matrices for tapered finite strip elements.
47
The shape function used in CUFSM is based on Cheung (1976). The shape
function is the multiplication of a width-direction shape function X(x) and a
longitudinal-direction shape function Y(y).
N(x,y) = X(x)Y(y) (27)
For the plane stress part of the stiffness matrix, the width-direction shape function
X(x) is linear, which is the same as a truss element. For the plate bending part of the
matrix, the width-direction shape function X(x) is cubic, which is the same as a beam
element. The longitudinal-direction shape function Y(y) is a sine function, since the
boundary condition at the loaded edges of the CUFSM program is limited to the
simply-supported case. Only the width-direction shape functions are modified, since
the thickness varies only in that direction. The exact shape functions for tapered truss
and beam elements are based on McGuire et al. (2000).
Figure 3.3 A tapered truss or beam element with unit width or a cross-section of
a tapered plate element
Figure 3.4 Degrees of freedom for a finite strip
48
For the axially loaded linearly tapered truss element shown in Figure 3.3, the
equilibrium of an infinitesimal segment in the x-direction yields the following
displacement field:
1 2
( ) ( )
T
t
u x N x u u = ( (
(28)
1 1
( ) 1 ( ) ( )
t
N x f x f x = ( (
(29)
where
( )
( )
1
ln 1
( )
ln 1
x
b
f x
+
,
2 1
1
t t
t
= and ( )
t
N x (
= the shape function for a tapered
truss element.
Using Equation (29), the shape function for the plane stress is expressed in the
form of Equation (27) within the displacement field:
| |
1 1 1 2 2
( , ) ( , ) ( , )
T T
u x y v x y N x y u v u v = ( (
(30)
| |
( )
( )
1 1
1
1 1
1 ( ) ( ) 0 ( ) ( ) 0
( , )
( ) ( )
0 1 ( ) 0 ( )
m m
m m
m m
f x Y y f x Y y
N x y
Y y Y y
f x f x
k k
(
(
=
(
(
(31)
where Y
m
(y) = sin(k
m
y), k
m
= m
/a, m = number of half-waviness, and a = longitudinal
length of the strip element (y-direction in Figure 3.4). The degrees of freedom in
Equation (30) are defined in Figure 3.4.
The strain field is derived from the displacement field:
| |
1 1 1 2 2
( , )
T
T
T
x y xy
u v u v
B x y u v u v
x y y x
( | | | | | |
( = + = (
( | | |
\ .
\ . \ .
(32)
| | ( )
( )
1 1
1 1 1
1 1 1 1
( ) ( ) 0 ( ) ( ) 0
( , ) 0 ( ) 1 ( ) 0 ( ) ( )
( ) ( )
1 ( ) ( ) ( ) ( ) ( ) ( )
m m
m m m m
m m
m m
m m
f x Y y f x Y y
B x y f x k Y y f x k Y y
Y y Y y
f x Y y f x f x Y y f x
k k
(
(
(
= (
(
(
(
(33)
49
The initial stiffness matrix of the plane stress part is derived from the
minimum potential energy:
[k
plane-stress
] =
1 1
0 0
( )[ ] [ ][ ]
b a
T
p
t x B E B dydx
(34)
where [E
p
] =
1 2
2 2
0
0
0 0
x
x
E E
E E
G
(
(
(
(
,
1
1
x
x y
E
E
=
,
2
1
y
x y
E
E
=
, ( )
1
1
x
t x t
b
| |
= +
|
\ .
, E
x
,
E
y
= Youngs moduli, G = shear modulus, and
x
,
y
= Poissons ratios.
For the linearly tapered beam element shown in Figure 3.3, the force
equilibrium of an infinitesimal segment in the vertical direction (z) and the moment
equilibrium with respect to the axis perpendicular to the cross-section (y) yield the
following displacement field:
1 1 2 2
( ) ( )
T
b
w x N x w w = ( (
(35)
where ( )
b
N x (
= the shape function for a tapered beam element
( )
( ) ( )
( )
( )
2 2
2
2 2
2
2
2
2
2 2
2
2
2
2 2
2
( ) ( )
( ) 2 ln
( ) ( )
ln(1 ) ln ( )
( )
1
( )
( )
( ) 2 ln ( )
( )
ln(1 )
( )
( ) ln ( )
( ) 2
b
f b f b
f x
f x f x
b x x
C b f x
f x b b
N x
C
f b
f x f x
f x
C
bf b x x
bf b f x
f x b b
| |
+ +
|
\ .
(
| | | |
+ + +
(
| |
\ . \ .
(
= (
+ + +
(
+ +
| |
+
(
|
\ .
(
T
`
)
(36)
where ( ) ( ) 2 2 ln 1 C = + + + , and
2
( ) 1
x
f x
b
= +
Using Equation (36), the shape function for plate bending is expressed in the
form of Equation (27) within the displacement field for a finite strip element:
2 1 1 2 2
( , ) ( , )
T
w x y N x y w w = ( (
(37)
50
2
( , ) ( ) ( )
b m
N x y N x Y y = ( (
(38)
The degrees of freedom in Equation (37) are defined in Figure 3.4.
Using the displacement field in Equation (37), the strain field is computed:
2 2 2
2 1 1 2 2
2 2
2 [ ( , )]
T
T
T
x y xy
w w w
B x y w w
x y x y
(
( = = (
(
(39)
where
| |
2
2
( , ) [ ( )] ( ) [ ( )] ( ) 2[ ( )] ( )
T
b m b m m b m
B x y N x Y y N x k Y y N x Y y ( =
.
The initial stiffness matrix of the plate bending part is also derived from the
minimum potential energy:
[k
plate-bending
] =
2 2
0 0
[ ] [ ][ ]
b a
T
b
B E B dydx
(40)
where [E
b
] =
3
( )
12
p
t x
E (
.
The complete initial stiffness matrix is a combination of Equations (34) and (40):
( )
( )
4 4
[ ]
4 4
plane stress
e
plate bending
k zeros
k
zeros k
(
(
(
=
(
(
(41)
The derivation of the geometric stiffness matrix follows Schafer (1997). The
thickness variation is also incorporated in the geometric stiffness matrix. For a linearly
varying applied stress,
( ) | | | |
1 1 2
0 0
( )
a b
T
g
x
k f f f t x G G dxdy
b
| |
( =
|
\ .
(42)
where
| |{ }
( , ) ( , ) ( , )
T
u x y v x y w x y
G d
y y y
(
=
(
, f
1
and f
2
= stresses at the nodes of
an element, {d}= u
1
v
1
u
2
v
2
w
1
1
w
2
2
T
, u(x,y) v(x,y)
T
= Equation (30), and
w(x,y) = Equation (37). The stiffness matrices derived in this manner are introduced
into the CUFSM program.
51
In this program, after the initial stiffness matrix of Equation (41) is computed
for each element, each stiffness matrix is transformed into a global coordinate system.
The transformed stiffness matrices from all elements are assembled according to the
global degrees of freedom. The same procedure is repeated for the geometric stiffness
matrix of Equation (42). By solving the eigen-value problem of Equation (43), elastic
buckling load factors () and corresponding mode shapes {} are obtained for a given
cross-section:
{ } { } 0
e g
K K ( + =
(43)
where K
e
, K
g
= the assembled initial and geometric stiffness matrices in a global
coordinate system, respectively.
Due to the singularity of the width-direction shape functions in Equations (29)
and (36) as the thickness variation ratio () approaches zero, the stiffness matrices for
a plate with tapered thickness are used only when the thickness variation ratio is larger
than or equal to 5%. For a thickness variation ratio less than 5%, stiffness matrices for
uniform thickness are used with an average thickness.
A version of the CUFSM program has been developed to accommodate the
effect of tapered thickness based upon the theory developed in this study. This version
of the program is called CUFSM-tap hereafter.
3.3 The Plate Buckling Coefficient for a Tapered Thickness Plate Element and
Application to the AA Specification
In the AA Specification (2000a), the compressive limit state stress varies
according to the equivalent slenderness ratio (
p
), which falls into one of three ranges,
the yielding, inelastic buckling, and post buckling ranges, as shown in Figure 2.4. The
equivalent slenderness ratio (
p
) is defined in Equation (5).
52
( )
2
12 1
, where
p
p
b
t k
| |
= =
|
\ .
(5)
As seen in this equation, the equivalent slenderness ratio of each component element
is dependent on the plate buckling coefficient (k
p
).
However, for a plate element with tapered thickness, the equivalent slenderness
is not provided in the current AA Specification (2000a), since the plate buckling
coefficient is unknown. Examples of sections with tapered thickness are shown in
Table 3.1. The shaded flange area of each cross-section is considered as an individual
plate element under uniform compression with idealized boundary conditions.
Table 3.1 Idealization of sections with tapered elements
Flexural member idealized
tapered element under uniform
compression is shaded
idealized boundary
condition (IBC)
IBC designation
SSSS
SSFS
SSSF
Since the thickness varies at the loaded edges, the distributed force per unit
length shown in this table, which is the multiplication of the uniform compressive
stress by the thickness, is not uniform. Using the CUFSM-tap program, the plate
buckling coefficient is computed for each idealized plate element shown in Table 3.1.
The computation is based on a given thickness variation ratio (), which is defined in
Equation (44), and repeated for different ratios. The thickness variation ratio () is an
additional variable needed to evaluate a plate with tapered thickness:
( )
2 1 1
t t t = (44)
53
The plate buckling coefficient (
1
p
k ) is computed based on the smallest thickness (t
1
).
( )
( )
2
2
1 1
2
12(1 ) /
cr
p
b t F
k
E
= (45)
The computed plate buckling coefficients are plotted in Figure 3.5. In the analysis, the
plate is discretized into eight equally spaced strips.
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
0
5
10
15
p
l
a
t
e
b
u
c
k
l
i
n
g
c
o
e
f
f
i
c
i
e
n
t
(
k
p
)
[SSSS]
CUFSM-tap (a=b)
Mizusawa (a=b)
(a)
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
0
1
2
3
4
5
p
l
a
t
e
b
u
c
k
l
i
n
g
c
o
e
f
f
i
c
i
e
n
t
(
k
p
)
[SSFS]
CUFSM-tap (a=b)
Mizusawa (a=b)
CUFSM-tap (a=200b)
(b)
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
0
2
4
6
8
p
l
a
t
e
b
u
c
k
l
i
n
g
c
o
e
f
f
i
c
i
e
n
t
(
k
p
)
[SSSF]
CUFSM-tap (a=b)
Mizusawa (a=b)
CUFSM-tap (a=200b)
(c)
Note: b/t
1
= 1000 in Mizusawa (1993)
Figure 3.5 Plate buckling coefficient of tapered thickness plates for boundary
condition (a) SSSS (b) SSFS (c) SSSF
1 2
curve-fit
0.739 4 4
p
k = + +
1 2
curve-fit
0.146 0.317 0.405
p
k = + +
1 2
curve-fit
0.335 0.480 0.405
p
k = + +
1
p
k
1
p
k
1
p
k
54
The plate buckling coefficients calculated in this study are compared to
Mizusawa (1993) as shown in Figure 3.5. Since the available data from Mizusawa
(1993) is limited to the case where the width (b) is the same as the length (a), initial
comparisons are made from buckling coefficients calculated for the same case.
However, since the plate buckling coefficient decreases as the length-to-width ratio
(a/b) increases under SSFS and SSSF boundary conditions, the results for a large
length-to-width ratio (a/b) of 200 are also given. The plate buckling coefficient
obtained from CUFSM by using this length-to-width ratio for a plate with uniform
thickness is 0.405. This number is rather small compared to 0.425 or 0.43, which is
found in other specifications. However, 0.405 is more consistent with the current AA
Specification (2000a), since the equivalent slenderness of a uniform thickness plate
under the SSFS (or SSSF) boundary condition is 5.13(b/t) according to the Aluminum
Company of America (1958), from which the basis for the AA Specification (2000a) is
obtained. See Equation (5) to obtain
p
= 5.13(b/t) from k
p
= 0.405 when = 1/3.
As the thickness variation ratio () increases, the plate buckling coefficient (k
p
1
)
increases monotonically as seen in Figure 3.5. For this reason, quadratic equations are
used for curve-fitting the relationship between the plate buckling coefficient and the
thickness variation ratio. The error average between the computed and curve-fitted
data is less than 0.1%, with a standard deviation of less than 0.5% for all cases.
The curve-fitted plate buckling coefficients in Figure 3.5 are based on the
minimum thickness (t
1
) as shown in Equation (45). However, it would be more useful
for practical design purposes if the plate buckling coefficients were expressed in terms
of the average thickness, as the current AA Specification is based on the average
thickness. The relationship between the plate buckling coefficients based on the
minimum and average thicknesses is obtained as follows:
( )
2
1
4 2
p p
k k = + (46)
55
where k
p
= the plate buckling coefficient based on the average thickness (t
avg
).
Equation (46) is derived from the following relationships:
( )( ) ( )( )
2 1 2
2 2
2
2
1
12 1
12 1
p p
cr
avg
k E k E
F
b t
b t
= =
and
1 2 2 1
1 1
1 2
2 2 2
avg
t
t t t t
t t
+ + +
= = = .
A comparison between the plate buckling coefficients of a uniform thickness plate and
a tapered thickness plate using Equation (46) is shown in Figure 3.6.
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
3.6
3.8
4
4.2
4.4
4.6
(a)
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
0.2
0.3
0.4
0.5
0.6
0.7
0.8
(b)
Figure 3.6 Rigorous plate buckling coefficient comparisons between tapered and
uniform thickness plates based on average thickness with (a) SSSS (b) SSFS and
SSSF boundary conditions
SSSS tapered thickness
p
k
uniform thickness (AA)
k
p
= 4.00
( )
( )
2
2
4 0.739 4 4
2
p
k
+ +
=
+
0 2.0
SSSF tapered thickness
SSFS tapered thickness
p
k
uniform thickness (AA)
k
p
= 0.405
( )
( )
2
2
4 0.335 0.480 0.405
2
p
k
+ +
=
+
( )
( )
2
2
4 0.146 0.317 0.405
2
p
k
+ +
=
+
0 2.0
56
The rigorous plate buckling coefficient for tapered thickness plates increases
as the thickness variation ratio () increases in the SSFS boundary condition case,
which implies that the current AA Specification is conservative. However, for other
cases, the current AA Specification is not conservative.
The equivalent slenderness ratios of tapered thickness plate elements are
obtained by introducing the rigorous plate buckling coefficients in Figure 3.6 into
Equation (5), as shown in Table 3.2. The rigorous equivalent slenderness ratios are
also linearly approximated in the table for design purposes. The errors due to linear
approximations of the rigorous equivalent slenderness ratios are within 2%. It is
noted that since the curve-fitted equations are for a limited range, the range of the
thickness variation ratio () is limited as well. However, the thickness variation ratios
of all standard I-shaped sections with tapered thickness in AA (2000b) fall within this
range.
Table 3.2 Linear approximation of the equivalent slenderness ratio (
p
) for
tapered thickness elements under uniform compression (0 < 2.0)
type of
member
rigorous expression
( )
p
linear approx.
( )
L
p
AA Spec.
( )
AA
p
[SSSS]
( )
( )
2
1.633 2
0.739 4 1 avg
b
t
| |
+
|
|
+ +
\ .
( ) 1.63 0.03
avg
b
t
| |
+
|
|
\ .
1.6
avg
b
t
| |
|
|
\ .
[SSFS]
( )
2
1.633 2
0.335 0.480 0.405 avg
b
t
| |
+
|
|
+ +
\ .
( ) 5.1 0.6
avg
b
t
| |
|
|
\ .
5.1
avg
b
t
| |
|
|
\ .
[SSSF]
( )
2
1.633 2
0.146 0.317 0.405 avg
b
t
| |
+
|
|
+ +
\ .
5.2
avg
b
t
| |
|
|
\ .
5.1
avg
b
t
| |
|
|
\ .
The limit state stress for a tapered flange element is determined by introducing
the linearly approximated equivalent slenderness ratios in Table 3.2 into the equations
in Table 3.3. The shape factors () in this table under the yield and ultimate limit
states are provided in Table 2.5b.
57
Table 3.3 Limit state stress equations for a tapered flange element
approaches
limit state stress
1
avg
b
S
t
limit
S
1
limit state stress
1 2
avg
b
S S
t
<
limit
S
2
limit state stress
2
avg
b
S
t
<
AA
Specification
AA
p cy
F F =
cy
AA
B F
D
AA AA
p p
F B D =
1
AA
k B
D
2 AA
p
AA
p
k BE
F
=
proposed linear
approximation
L
p cy
F F =
cy
L
B F
D
L L
p p
F B D =
1
L
k B
D
2 L
p
L
p
k BE
F
=
Note: a.
( )
AA AA
p avg
b t = and
( )
L L
p avg
b t = .
b. See Table 2.5b for .
0 5 10 15 20 25 30 35
0
0.2
0.4
0.6
0.8
1
(a)
0 5 10 15 20 25 30 35
-5
0
5
10
15
(b)
Figure 3.7 (a) The proposed limit state stress
( )
L
p
F and current limit state
stress
( )
AA
p
F for linearly tapered elements under uniform compression and
(b) error between the stresses ( = 1)
L AA
p p
L
p
F F
error
F
e
r
r
o
r
(
%
)
F
p
/
F
c
y
L
p
F
AA
p
F
= 1
6061-T6 (extrusion)
yield limit state
[SSFS]
[SSFS]
avg
b t
avg
b t
[SSSF]
[SSSS]
= 1
6061-T6 (extrusion)
yield limit state
58
For a given material (extruded 6061-T6) and geometry ( = 1), limit state
stresses (F
p
) from both approaches in Table 3.3 are plotted in Figure 3.7a for the SSFS
boundary condition. They confirm that the current AA Specification (2000a) is
conservative for the SSFS boundary condition. The errors between both approaches
are plotted in Figure 3.7b for all three boundary condition cases. The current AA
Specification (2000a) is unconservative for the SSSS and SSSF boundary conditions,
but the calculated errors for these boundary conditions are smaller than those for the
SSFS boundary condition.
3.4 The Moment Capacity Evaluation Approaches
The moment capacity evaluation approaches shown in Section 2.6 can be used
for the symmetric tapered thickness sections in Figure 3.1. However, the definition of
h
c
used in this section should be changed from the distance between centerlines to the
distance between centroids of the flange areas (A
f
) as shown in Figure 3.8.
Figure 3.8 Contributions of component elements to the entire moment capacity of
an I-shaped section with tapered thickness
3.5 Parametric Study of I-Shaped Sections with Tapered Thickness
The moment capacities are computed for all 36 standard I-shaped sections with
tapered thickness listed in AA (2000b) using the approaches developed for the tapered
59
thickness element. Subsequently the same sections are analyzed using the finite
element method. These sections are denoted by Series 1 in this parametric study. Since
the slenderness factors, defined in Equation (22), of all Series 1 sections are in the
stocky range ( < 0.673) based on AISI (1996), additional series sections are created
through reduction of the thickness of the standard sections by 60%. These sections are
denoted by Series 2. Details of the dimensions for Series 1 and 2 sections are given in
Section 6.4 of the Appendix.
The finite element model used for the parametric study is quite similar to that
of Section 2.7 except for the following. Each member length is at least four times the
member depth so that the end effects disappear at the middle of the member. The
length is also set to 3 to 8 times the critical local buckling length, which results in the
minimum buckling load. Twenty-noded quadratic hexahedral solid elements with
reduced integration are used to fully take into account the thickness variation in the
finite element model. The cross-sections used in the parametric study are extrusions of
6061-T6 with the minimum material properties in Table 2.4. Since the moment
capacity variation resulting from the ultimate strain variation is not significant, as
shown in Table 2.8, the ultimate strain is set to 8%, which is the same as the minimum
percent elongation in AA (2000c).
The influence of employing the rigorous plate buckling coefficient for a
tapered thickness plate is investigated, as shown in Figure 3.9. The abbreviations used
in Figure 3.9 and Figure 3.10 are as follows. In the first category, M
AA-UNI
and M
AA-
TAP
denote the moment capacities when the limit state stresses are computed by the
specification approach based on the uniform average thickness (F
p
AA
in Table 3.3) and
the proposed approach based on the tapered thickness (F
p
L
in Table 3.3), respectively.
The second category is the limit state, either yield (-
Y
) or ultimate (-
U
), as given in
Table 2.5 and Table 3.3 for the compression side. The third category indicates the
moment capacity evaluation approach in Table 2.7 with a correction for h
c
as
60
described in Section 3.4. On the vertical axis, the moment capacity from each finite
element analysis is divided by the capacity obtained from the approach specified in the
figure. As seen in Figure 3.9, considering the tapered thickness plate buckling
coefficient significantly improves the member capacity for rather slender sections
(Series 2 sections). This is because most Series 2 sections fall into the inelastic or post
buckling range, in which the limit state stress is considerably affected by the
equivalent slenderness ratio.
0.25 0.3 0.35 0.4 0.45 0.5 0.55 0.6
0.7
0.8
0.9
1
1.1
M
F
E
M
/
M
a
p
p
r
o
a
c
h
e
s
M
FEM
/ M
AA-UNI-U-WASA2
M
FEM
/ M
AA-TAP-U-WASA2
M
FEM
/ M
approaches
mean c.o.v
1.031 0.022
1.026 0.02
(a)
0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 1.6
0.6
0.8
1
1.2
1.4
M
F
E
M
/
M
a
p
p
r
o
a
c
h
e
s
M
FEM
/ M
AA-UNI-U-WASA2
M
FEM
/ M
AA-TAP-U-WASA2
M
FEM
/ M
approaches
mean c.o.v
1.198 0.044
1.102 0.052
(b)
Figure 3.9 Influence of employing the plate buckling coefficient for a tapered
thickness plate (a) Series 1 (b) Series 2
The two currently available approaches (M
AA-UNI-Y-MMCA
and M
AA-UNI-Y-WASA
)
in the AA Specification (2000a) are compared to the two proposed approaches (M
AA-
[Series 1]
36 AA standard
tapered I-sections
[Series 2]
36 tapered I-sections
with 60% thickness
reduction
y cr
= F F
y cr
= F F
61
TAP-U-WASA2
and M
AA-TAP-U-TMCA
). The comparison is shown in Figure 3.10. The two
proposed approaches predict the ultimate-plastic capacities of the cross-sections used
in the parametric study more accurately than the two currently available approaches
across a wide range of slenderness. In addition, the two proposed approaches show
similar results. The moment capacities obtained with all four approaches and the finite
element analyses normalized by the yield moment capacity are listed in Section 6.4.
0.25 0.3 0.35 0.4 0.45 0.5 0.55 0.6
0.4
0.6
0.8
1
1.2
1.4
M
F
E
M
/
M
a
p
p
r
o
a
c
h
e
s
M
FEM
/ M
AA-UNI-Y-MMCA
M
FEM
/ M
AA-UNI-Y-WASA
M
FEM
/ M
AA-TAP-U-WASA2
M
FEM
/ M
AA-TAP-U-TMCA
M
FEM
/ M
approaches
mean c.o.v
1.27 0.039
1.211 0.032
1.026 0.02
1.016 0.019
(a)
0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 1.6
0.5
1
1.5
M
F
E
M
/
M
a
p
p
r
o
a
c
h
e
s
M
FEM
/ M
AA-UNI-Y-MMCA
M
FEM
/ M
AA-UNI-Y-WASA
M
FEM
/ M
AA-TAP-U-WASA2
M
FEM
/ M
AA-TAP-U-TMCA
M
FEM
/ M
approaches
mean c.o.v
1.43 0.088
1.257 0.041
1.102 0.052
1.099 0.055
(b)
Figure 3.10 Comparison between current and proposed approaches (a) Series 1
(b) Series 2
[Series 1]
36 AA standard
tapered I-sections
[Series 2]
36 tapered I-sections
with thickness
reduction by 60%
I-3x1.96
y cr
= F F
y cr
= F F
62
3.6 Experiments and FEM Simulation
Flexural tests for three American Standard I-Beams with tapered thickness, I-
3x1.96, listed in AA (2000b), have been conducted to further support the approaches
proposed in this study. The dimensions of the section are shown in Section 6.4 of the
Appendix. The test setup is quite similar to that used for the uniform thickness
sections in Figure 2.16. It is simulated using the finite element method. All modeling
issues are similar to those of the SOLID model in Section 2.8 except for different
lateral support spacing, as shown in Figure 3.11, and different material properties, as
shown in Table 6.27c in the Appendix.
Note: all dimensions are in mm and not to scale
Figure 3.11 Schematic test setup (I-3x1.96)
In contrast to the uniform thickness sections of Section 2.8, almost no ripples
were formed during the tests, as shown in Figure 3.12. The beam does not seem to be
sensitive to such local deformations. The first test performed was unsuccessful due to
a lack of stiffness at the lateral supports; large lateral S-shaped deformations occurred,
as shown in Figure 3.12b. For this reason, the first tested beam reached the peak too
early, as seen in Figure 3.13. The test setup was reinforced against lateral movements
for the following tests, resolving the problem with the first test. However, the last two
tests were discontinued at a vertical displacement between approximately 150mm and
170mm at the span center, due to the limitation of the test setup. Up to this stage, load
factor continually increases as displacement increases, as observed in Figure 3.13.
63
(a)
(b)
Figure 3.12 Residual deformation of tested specimens (I-3x1.96) (a) side (b) plan
0 20 40 60 80 100 120 140 160 180
0
0.2
0.4
0.6
0.8
1
1.2
1.4
DESCRIPTION MAX-LF DISP-XL
TEST 1 1.123 50.56
TEST 2 1.173 168.5
TEST 3 1.174 147.2
FEM 1.197 167.4
DISP-XL = Displacement at max LF
SCVD (Span Center Vertical Displacement, mm)
L
F
=
M
u
/
M
y
Figure 3.13 Load factor-displacement result comparison for -3x1.96
64
The load factor-displacement curves obtained from the last two physical tests
are closely matched by the finite element analysis, as shown in Figure 3.13. In the
analysis, the Von Mises stress reaches the ultimate stress failure criteria without
stability problems. At this stage, the peak load factor is determined, although the load
factor is still increasing.
In Table 3.4, the average of the maximum moment capacities from the last two
physical tests is compared with those from the approaches available in the current AA
Specification (2000a) and those from the approaches developed in this study. The
proposed approaches show closer agreement with the tests than the current
approaches. Thus, in addition to the parametric study, the test results further validate
the proposed approaches. The moment ratios from the tests in Table 3.4 are slightly
smaller than from the finite element analyses in Figure 3.10 and Table 3.5, which may
be due to the early discontinuation of the tests. Since the material properties used for
the finite element analyses are not from the minimum values in AA (2000c) but from
tensile tests in this study, the moment capacities based on the approaches in these
tables are recalculated.
Table 3.4 Comparison of test results to the available approaches
current approaches proposed approaches
TESTS
AA-UNI-Y-MMCA
M
M
TESTS
AA-UNI-Y-WASA
M
M
TESTS
AA-TAP-U-WASA2
M
M
TESTS
AA-TAP-U-TMCA
M
M
1.174 1.141 0.967 0.960
Table 3.5 Comparison of finite element simulation to the available approaches
current approaches proposed approaches
FEM-SIMULATION
AA-UNI-Y-MMCA
M
M
FEM-SIMULATION
AA-UNI-Y-WASA
M
M
FEM-SIMULATION
AA-TAP-U-WASA2
M
M
FEM-SIMULATION
AA-TAP-U-TMCA
M
M
1.198 1.164 0.986 0.979
65
3.7 Application to the AA Specification
The allowable stress equations based on the yield and ultimate limit states are
summarized for tapered thickness elements in Table 3.6. The equations given in this
table are consistent with those proposed in Section 2.9 for uniform thickness elements.
In addition, the governing allowable stress for sections with tapered thickness can also
be determined by either Procedure I or Procedure II given in Table 2.9. Although
Table 2.9 and Table 3.6 are based on the Allowable Stress Design, these tables can
also be applied to the Load and Resistance Factor Design by replacing the reciprocal
of each safety factor with the corresponding resistance factor.
Table 3.6 Proposed allowable stress equations for tapered thickness elements
under uniform compression (0 < 2.0) when sections are under bending
BC
allowable
stress
1
avg
b
S
t
limit
S
1
allowable stress
1 2
avg
b
S S
t
<
limit
S
2
allowable stress
2
avg
b
S
t
<
y cy
ay
y
F
F
n
=
( ) 1.63 0.03
y cy
B F
D
+
SSSS
u cy
au
u
F
F
n
=
( ) 1.63 0.03
n
y
u cy n
u
B F
D
+
( ) 1.63 0.03
ay au
avg
y
F F
b
B D
t
n
= =
+
`
)
( )
1
1.63 0.03
k B
D +
( )
2
1.63 0.03
ay au
y
avg
F F
k BE
b
n
t
= =
| |
+
|
|
\ .
y cy
ay
y
F
F
n
=
( ) 5.1 0.6
y cy
B F
D
SSFS
u cy
au
u
F
F
n
=
( ) 5.1 0.6
n
y
u cy n
u
B F
D
( )
1
5.1 0.6
ay au
y avg
F F
b
B D
n t
= =
`
)
( )
1
5.1 0.6
k B
D
( )
2
5.1 0.6
ay au
y
avg
F F
k BE
b
n
t
= =
| |
|
|
\ .
y cy
ay
y
F
F
n
=
5.2
y cy
B F
D
SSSF
u cy
au
u
F
F
n
=
5.2
n
y
u cy n
u
B F
D
1
5.2
ay au
y avg
F F
b
B D
n t
=
| |
= |
|
\ .
1
5.2
k B
D
2
5.2
ay au
y
avg
F F
k BE
b
n
t
= =
| |
|
|
\ .
Note. See Section 2.9 for details.
66
3.8 Conclusions
Initial and geometric stiffness matrices for a plate with tapered thickness are
derived for the finite strip analysis program, CUFSM. Using the CUFSM program
with the derived stiffness matrices, parametric studies are conducted to find the
relationship between the plate buckling coefficient and the thickness variation ratio.
Using the relationship thus obtained, limit state stress equations are proposed for
tapered thickness elements, while the framework of the AA Specification (2000a) is
maintained.
Another parametric study is conducted for I-sections with tapered thickness for
the verification of the proposed limit state stress equations. The moment capacity
evaluation approaches that are developed for uniform thickness sections are also
employed for the sections used in the parametric study. From the comparisons with the
finite element analyses, it has been found that the proposed approaches work much
better than the current specification approaches. Experimental study additionally
supports the proposed approaches.
The limit state stress equations developed for tapered thickness elements are
improved primarily within the inelastic and post buckling ranges. For this reason, the
proposed approaches could also be used to develop more slender cross-sections with
tapered thickness than the current standard cross-sections, which mostly fall within the
yielding range.
67
4. NUMERICAL SLENDERNESS APPROACH FOR COMPLEX
ALUMINUM EXTRUSIONS UNDER FLEXURAL LOADING
One of the most attractive aspects of aluminum as a structural material is that it
can be extruded. Through extrusion, the cross-section can be a wide range of shapes,
and combination of shapes. For example, extrusion is the best process for producing
the geometric shape shown in Figure 4.1b, in which additional component elements
are attached to the flanges of a simple I-shaped section. Sections containing such
component elements are common among extruded aluminum shapes and provide
additional functions such as screw chases and grooves.
(a) (b)
Note: All dimensions are in inches and not to scale.
Figure 4.1 (a) Simple I-shaped section (b) complex extrusion
0
1
2
3
4
1 10 100
Figure 4.2 Buckling analyses by CUFSM for simple and complex extrusions
(8, 0.23)
(30, 0.94)
M
c
r
/
M
y
half-wavelength (in.)
67
68
As shown in Figure 4.2, the local buckling capacity of a complex extrusion
(Figure 4.1b) is usually much larger than a simple extrusion (Figure 4.1a) due to the
additional component elements. However, the advantage of increased member
capacity is usually disregarded due to the difficulty of considering all possible
geometric variations in the specifications. For example, in order to calculate the
compressive allowable stress (or factored limit state stress) of the complex extrusion
in Figure 4.1b using the AA Specification (2000a), the geometry should be simplified
to the one shown in Figure 4.1a.
To consider all possible geometric variations of aluminum extrusions, a
numerical analysis is desirable. Thus, a new design approach based on a numerical
analysis is developed. In this approach, the framework of the current specification is
maintained, with only the equivalent slenderness ratio being determined by local
buckling analysis. Since the geometric shape does not limit the numerical buckling
analysis, practically all aluminum extrusions can be evaluated. In this approach,
rigorously obtained plate buckling stress values are used instead of the minimum plate
buckling coefficients implicit in the AA Specification (2000a).
In addition, since the current methodologies to compute moment capacities
based on the determined limit state stresses are over-simplified for unsymmetric
sections, more precise approaches are formulated. A shape factor for a rectangular
web element is also developed for unsymmetric sections. Results of physical tests and
non-linear finite element analyses are compared to the approaches developed in this
study.
69
4.1 Numerical Slenderness Approach for Component Elements
In the AA Specification (2000a), the limit state (or allowable) stress for a
compressive component element varies according to the equivalent slenderness ratio
(
p
), which falls into one of three ranges, the yielding, inelastic buckling, and post
buckling ranges, as shown in Figure 2.4. The equivalent slenderness ratio is defined in
Equation (5).
( )
2
12 1
, where
p
p
b
t k
| |
= =
|
\ .
(5)
In the AA Specification (2000a), boundaries between component elements are
idealized as simply-supported, for which plate buckling coefficients (k
p
) are generally
known. It has been shown in Chapters 2 and 3 that this idealization is satisfactory for
sections with simple geometry such as in Figure 2.1 and Figure 3.1. Although the AA
Specification (2000a) provides limit state stress equations for additional types of
cross-sections, they are insufficient to cover the wide range of possible geometric
shapes through extrusion.
The Numerical Slenderness Approach (abbreviated as NSA hereafter) is
developed to provide a design approach for complex extrusions, in which the
aforementioned boundary idealization is not applicable or appropriate. In this
approach the equivalent slenderness ratio is determined using a numerical buckling
analysis tool such as CUFSM. The CUFSM program developed by Schafer (1997) is
designed to determine the minimum elastic buckling stress and a corresponding mode
shape for a given length of a member. By varying the member length, several mode
shapes such as local, distortional, and lateral buckling can be detected. Examples of
the buckling analysis results are shown in Figure 4.2. A complete manual for CUFSM
can be found in Schafer (1997), although the name of the program has changed since
then from CUSTRIP to CUFSM.
70
An alternative form of the equation for the equivalent slenderness ratio based
on the numerically determined minimum buckling stress F
cr
is derived from the
relationship given in Equation (6):
p
cr
E
F
= (47)
Since the geometric shape is not limited to simple cases when performing a numerical
analysis, the equivalent slenderness ratio can be obtained for a wide range of possible
geometries. There is only one unique minimum buckling stress corresponding to a
given cross-section, as shown in Figure 4.2. Thus, the equivalent slenderness ratios for
all component elements are equal. In addition, in the use of the NSA there is no need
to idealize boundaries between component elements as simply-supported, which is
necessary in the AA Specification (2000a). Using Equation (47), the limit state stress
equations are provided in Table 4.1. The equations in this table are identical to those
in Figure 2.4 for the AA Specification (2000a), except for the expression for the
equivalent slenderness ratio.
Table 4.1 Limit state stress equations for a component element
Limit state stress for
p
S
1
S
1
Limit state stress for
S
1
p
S
2
S
2
Limit state stress for
S
2
p
F
p
=
F
cy
cy
B F
D
F
p
= B D
p
1
k B
D
2
p
p
k BE
F
=
Note: B, D, k
1
and k
2
are the factors provided in the AA Specification (2000a). These factors differ
for flanges and webs. F
p
= limit state stress for flange (F
f
) or web (F
w
). See Table 2.5b for shape
factors () of symmetric cross-sections. For unsymmetric sections, see Table 4.8c.
4.2 Rigorous Ultimate Shape Factor for Rectangular Web Elements with Neutral
Axis Not at Mid-Depth
As shown in Section 2.5, the cut-off of the AA Specification (2000a) seen in
Figure 2.4 can be raised up to the ultimate shape factor (
u
) times the yield stress to
71
take into account the ultimate limit state of bending members. In this study, the closed
form solution of the ultimate shape factor for rectangular web elements when the
neutral axis is located at mid-depth is obtained using analytic integration of the stress
distribution.
However, when the neutral axis is not at mid-depth, as is the case where the
section is not symmetric with respect to the bending axis, the shape factor for the web
element given in Equation (14) is not appropriate.
Figure 4.3 Stress distributions corresponding to yield and ultimate moment
capacities when neutral axis (N.A.) is not located at mid-depth
Figure 4.3 shows the strain and stress distributions of a rectangular web
element, of which the neutral axis is not at mid-depth. Based on Figure 4.3, the closed-
form ultimate moment capacity and shape factor of a rectangular section with an
arbitrary neutral axis is derived using an analytical approach similar to the one for
symmetric sections in Section 2.3:
( )
( )
( )( )
( )
2 2 1 2 1 2 2
3 3
2 2 2
1
2
3 500 2 1 500 2
n n n n
c t c t
NA c t
u n n
c y y
n f f n f f
h y f f
M b
E n F n F E
+ + + +
| |
+ + +
| | + +
|
= + +
|
| + +
\ .
\ .
(48)
( )
3
2
3
3 / 2 4
y
y
NA
NA
F b
h
M y h
h y
| |
= +
|
+
\ .
(49)
( )
( )
( )
( )
( )( )
( )
3 2 1 2 1 2 2
3 3
2 2 2
2 3 2
1
3 2
3 500 2 1 500 2
4 3
n n n n
c t c t
NA
u c t
u w n n
y y y
y c
NA
n f f n f f
h y
M f f
M E n F n F E
F h y h
+ + + +
| |
+ + +
+
+
|
= = = + +
| + +
+
\ .
(50)
72
where y
NA
is the distance from the neutral axis to the mid-depth.
w
denotes the
ultimate shape factor of a rectangular web element with a neutral axis not at mid-depth
(for an unsymmetric section). For a symmetric section, another denotation (
wo
) has
been used, as shown in Equation (14).
In Equation (50), either the compression or tension extreme fiber stress (f
c
, f
t
)
should be the ultimate stress. If the compressive extreme fiber stress and strain (f
c
,
c
)
are the ultimate stress and strain (F
u
,
u
), respectively, the tension side extreme fiber
stress (f
t
) must also be determined to use this equation. This can be done by the
following procedure. After the tension side extreme fiber strain (
t
) is computed by
similar triangles, the strain is introduced into the Ramberg-Osgood equation, Equation
(7), to determine the tension side extreme fiber stress (f
t
) using increasing trial values.
It should be noted that Equation (14) is a special case (y
NA
= 0,
c
=
t
=
u
and f
c
= f
t
=
F
u
) of Equation (50).
4.3 Simplified Ultimate Shape Factor for Rectangular Web Elements with Neutral
Axis Not at Mid-Depth
Since Equation (50) requires the aforementioned additional iterative steps to
find the tension side extreme fiber stress, a simplification of this process is necessary
for application to practical design. Thus, using Equation (50), the shape factors are
computed and curve-fitted with respect to the location of the neutral axis, as shown in
Figure 4.4, for 6061-T6 alloy and temper. The curve-fitted expression in the figure,
shown again in Equation (51), is simple and accurate for any alloy and temper if the
two coefficients a and m are known. In Equation (51),
wo
can be determined from
Equation (14).
0.5 , 0 0.5
m
NA NA NA
u w wo
y y y
a
h h h
| || | | |
= = +
| | |
\ .\ . \ .
(51)
73
0 0.1 0.2 0.3 0.4 0.5
1.6
1.65
1.7
1.75
1.8
Curve Fitting for 6061-T6
a = 27.09
b = 1.597
m = 3.000
Figure 4.4 Ultimate shape factor for web with neutral axis not at mid-depth
(Individual Curve-Fitting)
Table 4.2 Material properties and coefficients of individual curve-fitting for
ultimate shape factor of solid rectangular sections for some 6000 series alloys
alloy-temper
a
F
y
(MPa)
b
E
(MPa)
a
F
u
(MPa)
c
u
e
a
d
wo
e
m
6005-T5, 6105-T5, 6351-T5 &
6061-T6, -T6510, & -T6511
241.15 68900 261.82 0.04 27.094 1.5975 3.00
6063-T5
f
110.24 68900 151.58 0.04 30.563 1.9467 3.03
6063-T5
g
103.35 68900 144.69 0.04 31.087 1.9766 3.04
6063-T6, T62 & 6463-T6 172.25 68900 206.70 0.04 28.423 1.7361 3.01
6066-T6, -T6510, & -T6511 310.05 68900 344.50 0.04 28.167 1.6224 3.05
6070-T6, -T62 310.05 68900 330.72 0.03 28.232 1.5617 3.09
Note: a. Minimum values from AA (2000c).
b. These are average values, which are 689 MPa (100 ksi) lower than compression.
c.
u
is assumed to be half of the minimum percent elongation listed in AA (2000c).
d. See Equation (14).
e. Coefficients obtained for Equation (51) using Individual Curve-Fitting.
f. Test coupon diameter or thickness up through 12.7mm.
g. Test coupon diameter or thickness between 12.7 and 25.4mm.
To determine the two coefficients in Equation (51), an iterative procedure is
employed until the best curve-fitting is obtained. This procedure is denoted by
Individual Curve-Fitting hereafter. For each 6000 series alloy listed in Table 4.2, the
coefficients are computed individually through this procedure. It is possible to tabulate
these coefficients for use in design. However, because of the continued increase in the
| || |
= +
| |
\ .\ .
0.5
m
NA NA
w wo
y y
a
h h
computed shape factor
h
y
NA
h/2
N.A
curve-fitted shape factor
w
y
NA
/h
6061-T6
a = 27.09
wo
= 1.598
m = 3.000
74
variety of alloy-temper combinations, a simpler method is desired. Thus, an
alternative approach is proposed, using the results from the Individual Curve-Fitting
method. The two coefficients are first determined for the same 986 alloy-temper
combinations in AA (2000c) used to obtain Equation (16). The results are plotted in
Figure 4.5 with respect to the ultimate stress to yield stress ratio. The results are
curve-fitted with rounded numbers for practical design purposes. This approach is
denoted by Unified Curve-Fitting hereafter.
14 13
tu
ty
F
a
F
= + (52)
m = 3 (53)
1 1.5 2 2.5 3 3.5
25
30
35
40
45
50
55
60
65
1 1.5 2 2.5 3 3.5
2
2.5
3
3.5
4
4.5
5
Figure 4.5 Unified Curve-Fitting parameters a and m
Introducing Equations (16), (52), and (53) into Equation (51), the final form of
the shape factor is obtained using the Unified Curve-Fitting:
3
14 13 0.5 1.25 0.2 , 0 0.5
tu NA NA tu NA
u w
ty ty
F y y F y
F h h F h
| | | |
| || | | |
= = + + +
| |
| | |
| |
\ .\ . \ .
\ . \ .
(54)
In Figure 4.6, the shape factor obtained from the Individual Curve-Fitting for each
material is divided by that from the Unified Curve-Fitting to investigate the accuracy
a
14 13
tu
ty
F
a
F
= +
Unified Curve-Fitting
tu ty
F F
Individual Curve-Fitting
m
Unified Curve-Fitting
m = 3
tu ty
F F
Individual Curve-Fitting
75
of the latter. Since the shape factor changes as the location of the neutral axis changes,
the range of the neutral axis variation (0 y
NA
/h 0.5) is divided into 20 equal
intervals. In this figure, the two solid dots connected by a solid line represent the
maximum and minimum ratios for each alloy-temper combination. Most of the shape
factors obtained from the Unified Curve-Fitting approximate those from the Individual
Curve-Fitting reasonably well. Thus, Equation (54) based on the Unified Curve-
Fitting is suggested for the AA Specification (2000a). For the complete NSA limit
state stress equations, the ultimate shape factor for a web element (
u
) implicit in
Table 4.1, which is Equation (16), should be replaced with that represented by
Equation (54).
1 1.5 2 2.5 3 3.5 4
0.5
0.6
0.7
0.8
0.9
1
1.1
1.2
1.3
1.4
1.5
Figure 4.6 Comparison between shape factors from Individual Curve-Fitting and
Unified Curve-Fitting
4.4 Moment Capacity Evaluation Approaches
Once the allowable stresses (or factored limit state stresses) for component
elements are obtained in actual design, the moment capacity can be computed using
one of the following approaches. However, instead of the allowable stresses (or
=
ICF (Individual Curve-Fitting) Eqn.(51)
UCF (Unified Curve-Fitting) Eqn.(54)
Note:
a. Orientation of specimen: no transverse direction
b.
u
> 1.5% (
u
0.5 percent elongation)
c. Total number of alloy-temper combinations = 986
d. Mean = 0.995, standard dev. = 0.018, min.=0.907
tu ty
F F
ICF
UCF
76
factored limit state stresses), limit state stress values have been used in the parametric
study that follows in Section 4.5.
4.4.1 Minimum Moment Capacity Approach (MMCA)
The MMCA is one approach described in AA (2000b), as explained in Section
2.6. However, for sections unsymmetric with respect to the bending axis, the
calculation sequence is slightly different from the symmetric case. First, the allowable
stress (or factored limit state stress) from each component element is multiplied by the
elastic section modulus of the entire cross-section to compute the moment capacity.
Among the obtained moment capacities, the minimum value is chosen as the
allowable moment capacity of the member. In this computation, the section modulus
of the entire cross-section is not uniform among component elements. This is because
the section modulus is the moment of inertia of the entire cross-section (I) divided by
the distance from the neutral axis to the location where the allowable stress (or
factored limit state stress) of each component element is evaluated. For the cross-
section given in Figure 4.7, the allowable moment capacity is
M
u
= min(F
cf
I/c
cf
, F
tf
I/c
tf
, F
cw
I/c
cw
, F
tw
I/c
tw
, F
cs
I/c
cs
) (55)
where F
cf
, F
tf
, F
cw
, F
tw
, and F
cs
are the limit state (or allowable) stress of compression
flange, tension flange, compression web, tension web, and compression edge-stiffener,
respectively. Other notations are given in Figure 4.7. This approach is demonstrated in
AA (2000b) through the Illustrative Examples of Design (Part VIII, Example 22) for a
similar type of unsymmetric cross-section.
4.4.2 Weighted Average Stress Approach (WASA)
The WASA is an alternative approach in the AA Specification (2000a) to
compute the moment capacity, as explained in Section 2.6. For symmetric sections
with edge-stiffeners such as C or Z-sections (Figure 4.10c), additional terms for the
77
edge-stiffener are added, as shown in Table 4.3. The modification factor (h/h
c
) is also
recommended for this type of cross-section as is done in Section 2.6.
Table 4.3 Correction in WASA for symmetric edge-stiffened C or Z sections
current WASA (WASA) proposed WASA (WASA2)
1
3
1
3
f f w w s s
u
f w s
F A F A F A
M S
A A A
+ +
=
+ +
1
3
1
3
f f w w s s
u
f w s c
F A F A F A
h
M S
A A A h
+ + | |
=
|
+ +
\ .
where F
f
, F
w
, F
s
= limit state (allowable) stress for flange, web, and stiffener,
respectively; A
f
= entire compression side flange area, A
w
= half of web area,
A
s
= entire compression side stiffener area, S = section modulus = I
/(h/2),
h = entire depth of a section, and h
c
= depth of a section measured between
the centroids of compression and tension flanges.
However, for sections with a neutral axis not at the mid-depth (or unsymmetric
sections), this approach is not applicable because the simplification procedure used in
Equation (19) is based on the assumption that the cross-section is symmetric so the
weighted average stress cannot be calculated using Equation (18).
4.4.3 Total Moment Capacity Approach (TMCA)
The TMCA introduced in Section 2.6 can be extended to most possible cross-
sectional geometries. For example, the sections shown in Figure 4.10 can be divided
into the web group (shaded area) and the flange group (unshaded area). The first three
cross-sections have the elastic neutral axis at the mid-depth, while the remaining two
cross-sections do not.
For cross-sections with a neutral axis located far from mid-depth, the use of a
single linear stress distribution for both tension and compression flanges would not be
realistic. For example, in Figure 4.7 the tension limit state stress (F
tf
) obtained from a
linear stress distribution based on the compression limit state stress (F
cf
) could cause a
significant difference from the actual tension limit state stress (F
tf
). For this reason,
the flange group may be subdivided into two groups, resulting in a total of three
groups in an entire cross-section, i.e., the compression flange group, tension flange
group, and web group. This approach is denoted by TMCA2. The moment capacity
78
based on TMCA2 is expressed in Equation (56) using the notations given in Figure
4.7.
M
u
= M
cf
+ M
tf
+ M
w
= F
cf
S
cf
+ F
tf
S
tf
+ F
cw
S
w
=
cf cf tf tf
cw w
cf tf cw
F I F I
F I
c c c
+ + (56)
When calculating the moment of inertia for each group, it is recommended that
the simple groups be computed first, e.g., I
cf
and I
w
in Figure 4.7. Then the remaining
moment of inertia for the complicated group (I
tf
in this figure) can be obtained by
subtracting the sum of the previously calculated groups from the moment of inertia of
the entire cross-section (I), which is generally provided by the manufacturer:
I
tf
= I (I
cf
+ I
w
) (57)
Figure 4.7 Linear approximations of an actual non-linear stress distribution of a
mullion section
In this example the moment capacity could not have been computed properly
with the WASA2, due to the complex tension flange and the location of the neutral
axis. In contrast, Equations (56) and (57) show the versatility of the TMCA2 in
tackling complex cross-sections, although the following additional steps are required
to compute the limit state stress of the tension flange. First, the stain corresponding to
79
the compressive limit state stress is computed using the Ramberg-Osgood equation
shown in Equation (7). A linear strain distribution is then drawn from the calculated
strain through the neutral axis to the tension extreme fiber. The strain thus determined
at the tension extreme fiber is introduced into the Ramberg-Osgood equation to obtain
the limit state stress for the tension flange (extreme fiber) by an iterative process. For
either the flange or web stress distribution, the tension side extreme fiber stress
(tension limit state stress) must be limited to a maximum value of the corresponding
shape factor times the yield stress. This limitation of the maximum extreme fiber
stress is not necessary for the case of symmetric sections because the symmetry itself
imposes limitations.
TMCA2 becomes the same as TMCA when the cross-section is symmetric
with respect to the bending axis. In addition, TMCA2 becomes TMCA as the
slenderness increases, because the non-linear stress distribution becomes linear due to
a loss of the inelastic reserve capacity. In the use of TMCA2, the stress of each
component element given in Equation (56) (F
cf
, F
tf
, and F
cw
) should not be the
allowable stress (or factored limit state stress) but the limit state stress, for the
following reason. If each limit state stress is first divided by the safety factor (or
multiplied by the resistance factor), the non-linearity of the stress distribution
decreases, resulting in almost no difference between TMCA and TMCA2. Therefore,
the moment capacity is computed using the limit state stresses after which the entire
moment capacity should be divided by the safety factor (or multiplied by the
resistance factor). See Table 6.19 in the Appendix for a step-by-step procedure.
4.4.4 Moment Capacity Based on Elasto-Plastic Stress Distribution (EPMC)
An approach to computing the moment capacity using the elasto-plastic stress
distribution is available in AISI (1996). In AISI (1996), the ultimate compressive
80
strain is calculated with a factor proposed by Reck, Pekz, and Winter (1975) based
on the slenderness of the compression flange.
However, the use of this factor in calculating the ultimate compressive strain is
not necessary in the present study. Instead, the ultimate compressive strain can be
directly determined from the Ramberg-Osgood equation, Equation (7), since the
compressive limit state stress for the flange element is known based on either the NSA
or the AA Specification (2000a). The strain distribution is determined based on the
compressive limit state strain and an assumed neutral axis location, which is set as an
unknown variable. The stress distribution is determined from the linear stress-strain
relationship, unless the strain is larger than the yield strain, in which case it is assumed
to be equal to the yield stress. The neutral axis location is computed from the force
equilibrium with respect to the longitudinal direction. The moment capacity is then
computed on an element-by-element basis. Since this approach uses the elasto-plastic
stress distribution without strain hardening, it is expected to be somewhat conservative
for some compact sections.
Although this approach seems straightforward, it becomes complicated when
there are additional elements between compression and tension flanges, such as edge-
stiffeners. This is because depending on the location of the neutral axis the stress
distribution at the edge-stiffener changes. For each possible case, equilibrium of the
stress distribution is used to determine the neutral axis, which is then compared with
the initially assumed location. For the edge-stiffened singly-symmetric section shown
in Figure 4.10d, a total of seven cases must be evaluated to find the neutral axis
location, as shown in Figure 4.8. However, for different cross-sections, different cases
must be evaluated, limiting the practicality of this approach.
81
Note: strain zone that is not yielded is shaded.
Figure 4.8 Possible cases using EPMC
4.4.5 Moment Capacity Based on Ramberg-Osgood Stress Distribution
(ROMC)
The ROMC uses the Ramberg-Osgood equation, Equation (7), to determine the
stress distribution from the strain distribution. The stress distribution obtained is
analytically integrated to compute the moment capacity. In the sense that the approach
computes the location of the neutral axis from the equilibrium and relies on a linear
strain distribution, it is similar to the EPMC. The following describes the overall
procedure of this approach, based on an example given in Figure 4.9.
Depending on whether the tension or compression yields first, two possible
stress distribution cases are considered, as shown in Figure 4.9. For either stress
distribution case, the compression limit state stress (f
c
) is computed first. Using the
Ramberg-Osgood equation, the corresponding compression strain (
c
) is obtained from
Equation (60). Based on the strain and the unknown distance between the mid-depth
and the neutral axis (y
NA
), a linear strain distribution is assumed. The tension side strain
(
t
) is determined from the strain distribution. Using the Ramberg-Osgood equation, a
82
relationship between the tension side strain (
t
) and the tension limit state stress (f
t
) is
obtained:
0.002
n
t t
t
y
f f
E F
| |
= +
|
|
\ .
(58)
(a) (b) (c)
Figure 4.9 Notations for edge-stiffened singly symmetric section (a) cross-section
(b) compression yield first (c) tension yield first
The assumed stress distribution needs to satisfy the force equilibrium with respect to
the longitudinal direction:
0
A
f dA =
(59)
Since there are two unknowns, f
t
and y
NA
, and two equations, Equations (58) and (59),
the unknowns can be solved. However, due to the highly non-linear nature of the
equations, the procedure for obtaining solutions is iterative.
In some cases of Figure 4.9c, the computed tension limit state stress (f
t
) could
be larger than the ultimate stress. In this case, the tension limit state stress (f
t
) should
be reduced to the ultimate stress. According to the reduced tension limit state stress
(f
t
), the entire strain stress distribution is reduced. The compression limit state stress is
then back calculated from this reduced strain distribution with the Ramberg-Osgood
equation:
0.002
n
c c
c
y
f f
E F
| |
= +
|
|
\ .
(60)
83
In this case, the unknowns are f
c
and y
NA
, which can be solved by two equations:
Equations (59) and (60).
After the stress distributions and the location of the neutral axis are
determined, a rather rigorous moment capacity can be obtained using classical beam
theory:
u
A
M f y dA =
(61)
or using the notations given in Figure 4.9,
3
2
c
u c f c
h
M f t w y
| |
= +
|
\ .
+ f
t
t
f
w
t
3
2
c
h
y
| |
|
\ .
2
1 3
3 3
2 2
2
2
t
f
s c
c s s h h
c c
d h
f t d z y d
y y
| |
| |
|
+ + +
|
|
\ . + +
\ .
+
( )( )
( )
( )
( )
2
2 2 2 1 2 1
3 3
3
2
2 2
1
3 500 2 500 2 1
h n n n n
c
c t c t
c t
w n n
t y y
n f f n f f
y
f f
t
E n F E n F
+ + + +
| | | | + + +
+
| | + +
| | + +
\ . \ .
(62)
where z
1
is the distance from the neutral axis to the centroid of the stress distribution
on the stiffener, which is assumed to be linear.
The ROMC should be the most accurate approach among those developed in
this study, since a non-linear stress distribution is used to compute the neutral axis,
rather than a linear elastic stress distribution. However, this approach is rather
complicated due to the iterative nature of the procedure. In addition, the equations for
finding the neutral axis location and moment capacity, Equation (62), need to be re-
derived for different types of cross-sections. Thus, this approach is not appropriate for
practical design.
84
4.5 Parametric Studies
4.5.1 Types of Analyses Used
The approaches to computing the moment capacity of an aluminum member
proposed in this study and presented in the AA Specification (2000a) are summarized
in Table 4.4.
Table 4.4 Types of analysis approaches (a) specification applicable (b)
specification non-applicable
(a)
AA Specification proposed design approaches
design categories
M
AA-Y-MMCA
M
AA-Y-WASA
M
NSA-U-WASA2
M
NSA-U-TMCA
M
NSA-U-TMCA2
limit state yielding yielding ultimate ultimate ultimate
equivalent
slenderness ratio
AA AA NSA NSA NSA
neutral axis LENA LENA LENA LENA LENA
moment capacity MMCA WASA WASA2 TMCA TMCA2
recommended
cross-section
symmetric or
unsymmetric
symmetric symmetric
symmetric or
unsymmetric
symmetric or
unsymmetric
(b)
proposed design approaches
design categories
M
NSA-U-EPMC
M
NSA-U-ROMC
limit state elasto-plastic ultimate
equivalent
slenderness ratio
NSA NSA
neutral axis computed from equilibrium computed from equilibrium
moment capacity EPMC (element-by-element basis) ROMC (Equation (62))
recommended
cross-section
symmetric or unsymmetric symmetric or unsymmetric
Note: LENA = a neutral axis determined from a linear elastic stress distribution.
MMCA = Minimum moment capacity approach in Section 4.4.1.
WASA = Weighted average stress approach in Section 4.4.2.
WASA2 = Modified weighted average stress approach in Section 4.4.2.
TMCA = Total moment capacity approach (2 groups) in Section 4.4.3.
TMCA2 = Total moment capacity approach (3 groups) in Section 4.4.3.
EPMC = Moment capacity based on elasto-plastic stress distribution in Section 4.4.4.
ROMC = Moment capacity based on Ramberg-Osgood stress distribution in Section 4.4.5.
85
The first three proposed approaches in this table are suited for application in design,
while the remaining two are somewhat too complicated. The first category in Table
4.4 defines the limit state used in each approach, which is described in Section 2.1.
The second category indicates whether the equivalent slenderness ratio (as well as the
compressive limit state stress) for each component element is determined using the
AA Specification (2000a, Section 2.2) or the Numerical Slenderness Approach (NSA,
Section 4.1). The third category represents whether the neutral axis is directly
calculated from equilibrium or determined based on the linear elastic stress
distribution (LENA). The final category indicates the method used to calculate the
moment capacity.
4.5.2 Cross-Sections Used in Parametric Study
A parametric study is performed for five types of sections, as shown in Figure
4.10. The first two sections are simple extrusions, which can be designed accurately
following the proposed approaches in Chapter 2 and 3. However, these sections are
included in this study so that the applicability of the NSA to the simple extrusions can
also be examined. The NSA is originally designed for complex extrusions, as
explained in Section 4.1.
(a) (b) (c) (d) (e)
Note: web group area is shaded.
Figure 4.10 Cross-sections used in the parametric study
Although there is a provision in the AA Specification (2000a) for the edge-
stiffened sections shown in Figure 4.10c and Figure 4.10d, unlike the more general
86
NSA, it does not deal with cross-sections with different flange and stiffener
thicknesses. In addition, it is limited to the yield limit state. There is no provision in
the AA Specification (2000a) for taking into account multiple intermediate stiffeners
and variation in thickness within one component element, as is the case for the cross-
section in Figure 4.10e.
The slenderness of the cross-sections used in the parametric study is rather
low, such that the equivalent slenderness ratios of component elements mostly fall into
the yielding or inelastic buckling ranges defined in Figure 2.4. This is because the
slenderness of most of the standard sections in AA (2000b) also falls within these
ranges. Fixed dimensions of the cross-sections are given below, and the varied
dimensions used in the parametric study can be found in Section 6.4 of the Appendix.
4.5.2.1 Symmetric I-Shaped Sections with Uniform Thickness (Figure 4.10a)
The cross-sections used in this parametric study are the same as in Section 2.7.
The depth between flange center-lines and the width of the flanges are maintained at
254mm (10 in.), while the thicknesses are varied. The length is set to ten times the
depth. The width-to-thickness ratio for the flange ranges from 2.70 to 9.75, and from
8.33 to 38.0 for the web. This covers most of the width-to-thickness ratios of the
standard doubly symmetric I-sections listed in AA (2000b).
4.5.2.2 Symmetric I-Shaped Sections with Tapered Thickness (Figure 4.10b)
The cross-sections used in this parametric study are the same as in Section 3.5.
All the 36 standard I-sections with tapered thickness listed in AA (2000b) are used for
Series 1. Since all of these sections are rather stocky, additional sections are created
through reduction of the thickness of the standard sections by 60% that constitute
Series 2. Width-to-average thickness ratio for the flange ranges from 3.5 to 21.6, while
width-to-thickness ratio for the web ranges from 6.6 to 81.5. Each member length is
set to more than or equal to four times the member depth.
87
4.5.2.3 Edge-Stiffened Z-Sections (Figure 4.10c)
Depth and flange width along the mid-thickness are maintained at 91.44mm
(3.6 in.) and 45.72mm (1.8 in.), respectively, while the thicknesses for each
component element are varied. Width-to-thickness ratio for the flange ranges from 2.2
to 21, and ranges from 6.3 to 43 for the web. In Series 1, thicknesses of all component
elements are the same. In Series 2 stiffeners are twice as thick as the other elements,
as shown in Figure 4.11. Since the dimensions of tension elements are identical to
those of compression elements and the sections are laterally supported, the behavior of
edge-stiffened Z-sections should be similar to the edge-stiffened C-sections. Thus, this
type of section is considered as symmetric when design approaches are applied.
Member length is taken as three times the half-wavelength corresponding to the
minimum distortional buckling stress mode obtained from buckling analyses.
(a) (b)
Note: All dimensions are in mm (and inches inside parenthesis) and not to scale.
Figure 4.11 Model geometry and dimensional notations for edge-stiffened Z-
sections (a) Series 1 (b) Series 2
4.5.2.4 Singly Symmetric Edge-Stiffened Sections (Figure 4.10d)
Two different series of cross-sections used in the parametric study are
investigated. For both series, the width of the compression flange is 228.6mm (9 in.),
88
the depth of the member is 279.4mm (11 in.), and the length of the stiffener is 63.5mm
(2.5 in.) when measured through mid-thickness. Tension flange width varies from 25.4
mm (1 in.) to 431.8mm (17 in.). The varying width is used to evaluate the influence of
the location of the neutral axis on each approach. The thickness of the flanges and web
is 25.4mm (1 in.) for one series and 7.62mm (0.3 in.) for the other. The thickness of
edge-stiffeners is set to twice the flange thickness. The length of the member is set to
three times the half-wavelength corresponding to the minimum distortional buckling
stress mode obtained from buckling analyses.
(a) (b)
Note: all dimensions are in mm (and inches inside parenthesis) and not to scale.
Figure 4.12 Model geometry and dimensional notations for edge-stiffened singly-
symmetric sections (a) Series 1 (b) Series 2
4.5.2.5 Dome-strut sections (Figure 4.10e)
There are two series of cross-sections used in this parametric study. In Series
2, the web-flange junction is protruded, while it is not protruded in Series 1, as shown
in Figure 4.13. The depth measured between mid-thicknesses of the flanges and the
89
width of flanges remain constant at 127 mm (5 in.). Interior intermediate stiffeners are
12.7 mm (0.5 in.) long, while exterior stiffeners are 10.16mm (0.4 in.) long. Width-to-
thickness ratio for the flange ranges from 4.1 to 27.2 if the intermediate stiffeners are
ignored. Width-to-thickness ratio for the web ranges from 12 to 86.4 if the portions
protruding from the web-flange junctions are ignored. The thickness of the thicker
portion of the web (t
w2
) is set to three times the thinner portion (t
w
). The length of the
thicker portion of the web (d
w
) is set to one-eighth of the section depth. All the
thicknesses are varied for this parametric study. The length of the member is set to
three times the half wavelength corresponding to the minimum distortional buckling
stress mode obtained from the buckling analyses.
(a) (b)
Note: all dimensions are in mm (and inches inside parenthesis) and not to scale.
Figure 4.13 Model geometry and dimensional notations for dome-strut sections
(a) Series 1 (b) Series 2
4.5.3 Finite Element Modeling
Second order inelastic non-linear finite element analyses are conducted using
ABAQUS, developed by Hibbitt, Karlsson & Sorensen, Inc. (1998), for comparison
90
with the design approaches proposed in this study and those presented in the AA
Specification (2000a) as summarized in Table 4.4.
For all cross-sections used in the parametric study except type (c) sections,
rigid beam elements are attached at the ends of the member so that a plane section can
remain plane during bending. For these sections, equal and opposite concentrated
moments are applied at both ends to simulate pure bending conditions. For type (c)
sections, consistent nodal forces are applied to simulate pure bending conditions.
Lateral supports are attached at web-flange junctions so that lateral buckling is
prevented. Initial geometric imperfections are generated using elastic buckling
analyses with the maximum amplitude based on the standard flatness tolerance
provided in AA (2000c).
For type (b) sections, twenty-noded quadratic hexahedral solid elements with
reduced integration are used to fully take into account the tapered thickness. For all
other types of sections, shell elements are used. Further details regarding finite
element modeling can be found in Section 2.5.
4.5.4 Material Properties
The cross-sections used in the parametric study are extrusions of 6061-T6 with
the minimum material properties tabulated in AA (2000c) and shown in Table 2.4.
Based on these properties, a tensile stress-strain curve is fitted by the Ramberg-
Osgood equation given in Equation (7). The stress-strain curve for compression is
assumed to be the same as for tension. The ultimate strain, which is not defined in AA
(neither 2000a, 2000b or 2000c), is set to 8%, based on the discussion set forth in
Section 2.3.
91
4.5.5 Idealization of Type (e) Sections in the AA Specification (2000a) and NSA
The dome-strut section shown in Figure 4.10e is one of the most common
types of extrusion. It consists of a doubly symmetric I-shaped portion and multiple
stiffeners on one of the flanges. The stiffeners are intended for use as a screw chase to
hold the exterior panels of a dome structure. The behavior of sections with (Figure
4.10e) and without stiffeners (Figure 4.10a) is significantly different, which is also
demonstrated in Figure 4.2. This difference affects the calculated member capacity.
Figure 4.14 Idealization in the AA
Specification (2000a) for a dome strut section
However, the AA Specification (2000a) does not have provisions for multiple
stiffeners as in this dome-strut section. In addition, the variation in the web thickness
near the web-flange junctions cannot be taken into account in the specification. A
possible idealization made for the limit state (or allowable) stress computation of
component elements using the AA Specification (2000a) is shown in Figure 4.14. In
contrast, the NSA can account for the irregularities of the section. However, the
thickness variation in the web is not considered when the shape factor is computed,
because it is based on uniform thickness as shown in Figure 4.3.
4.5.6 Results
The ultimate moment capacity obtained from each finite element analysis for
each section given in Figure 4.10 is divided by that from each approach in the AA
Specification (2000a) and proposed in this study. The results are shown in Figure 4.15
92
with respect to the slenderness factor (), as defined in Equation (22), in the horizontal
axis. Since AISI (1996) defines a component element as fully effective when is less
than or equal to 0.673, it would be reasonable to consider 0.673 as the border between
the slender and stocky ranges. However, the slenderness factor is not used in design
procedures but for visual convenience only.
For all sections, the proposed approaches using the NSA show a performance
superior to the approaches in the AA Specification (2000a). The differences amongst
the approaches are summarized in Table 4.4. For the stocky range ( 0.673), the
difference between the proposed and specification-based approaches should be mainly
due to the choice of limit state (ultimate or yield). In this range, proposed approaches
predict the ultimate-plastic capacities quite precisely for all type of sections, when
they are compared to the finite element analyses. On the other hand, for the slender
range ( > 0.673), the accuracy of the approach should depend more on the method
that the equivalent slenderness ratio is determined. For this reason, when the limit
state stress is computed based on highly idealized geometry as shown in Figure 4.14,
the bending capacity cannot be computed accurately using the current specification
approach as shown in Figure 4.15e; the moment capacity computed is at times less
than half of that computed using the FEM analysis for some cross-sections.
In addition, the proposed approaches based on the WASA2, TMCA or
TMCA2 show a closer agreement with the finite element analysis than the
specification approaches based on the MMCA or WASA. This supports the accuracy
of the proposed approaches. Especially for unsymmetric cross-sections, the MMCA is
the only specification-based approach that can be used as explained in Section 4.4.2.
Since contributions from all component elements cannot be fully incorporated in the
MMCA as opposed to the other aforementioned approaches, the deviation between
this approach and the finite element analysis is larger for unsymmetric cross-sections.
93
The following subsections further elaborate on the results of the parametric study for
each type of cross-section shown in Figure 4.10.
4.5.6.1 Symmetric I-Shaped Sections with Uniform Thickness (Figure 4.10a)
For this type of cross-section, the approaches using the NSA in Figure 4.15a
show almost the same performance as the proposed approaches in Figure 2.14. This
implies that the idealization of the web-flange junction as being simply-supported is
quite accurate for simple I-shaped sections with uniform thickness.
4.5.6.2 Symmetric I-Shaped Sections with Tapered Thickness (Figure 4.10b)
For use in the NSA, the buckling stress of the tapered thickness sections is
computed using CUFSM-tap, which is the version of the CUFSM program modified
for the tapered element based on the study in Section 3.2. For this type of cross-
section, the idealization of simply-supported boundaries between component elements
is quite accurate if the equivalent slenderness ratio developed in Section 3.3 for simple
I-shaped sections with tapered thickness is used, which is almost the same conclusion
made for the I-sections with uniform thickness.
4.5.6.3 Edge-Stiffened Z-Sections (Figure 4.10c)
For this type of cross-section, the results based on the AA Specification
(2000a) are not significantly different from those based on the NSA in the slender
range ( > 0.673). The approach available in the AA Specification (2000a) for edge-
stiffened sections seems reasonable.
4.5.6.4 Singly Symmetric Edge-Stiffened Sections (Figure 4.10d)
For the singly symmetric edge-stiffened section, it is noted that the moment
capacity based on the Ramberg-Osgood equation (ROMC) shows excellent agreement
with the non-linear finite element analysis as shown in Figure 4.15d. In this figure,
94
although the proposed approach based on the TMCA is still much better than that of
the AA Specification (2000a), the difference from the results of the finite element
analysis increases as the distance from the neutral axis to the mid-depth (y
NA
)
increases. On the other hand, the other proposed approach using the TMCA2 shows
considerable improvement compared to the approach using the TMCA, because the
TMCA2 uses three linearized stress distributions. Thus, the use of TMCA2 is
recommended for unsymmetric sections.
4.5.6.5 Dome-Strut Sections (Figure 4.10e)
In order to investigate the effect of the idealizations shown in Figure 4.14, type
(e) sections are analyzed at the yield limit state in Figure 4.16. As seen in this figure, a
more significant difference between the approach based on the AA Specification
(2000a) and the proposed approach is observed in the larger slenderness range, where
the limit state stress is more sensitive to the equivalent slenderness ratio. In addition,
Series 2 sections show more deviation from the finite element analysis than Series 1
sections due to the additional idealization in the web element. Thus, the equivalent
slenderness ratio computed through the NSA should be applied to the design of
complex extrusions.
Finally, the performance improvement from the use of both the ultimate limit
state and the TMCA2 is demonstrated in Figure 4.15e. Although the results for the
TMCA and TMCA2 are similar for large slenderness factors, as the slenderness of the
cross-section decreases the difference between these two methods becomes more
prominent. This is because as the slenderness increases the stress distribution becomes
more linear due to the lack of non-linear inelastic reserve capacity.
95
0.3 0.4 0.5 0.6 0.7 0.8
0
0.5
1
1.5
2
M
F
E
M
/
M
a
p
p
r
o
a
c
h
e
s
M
FEM
/ M
approaches
mean c.o.v
M
FEM
/ M
AA-Y-MMCA
M
FEM
/ M
AA-Y-WASA
M
FEM
/ M
NSA-U-WASA2
M
FEM
/ M
NSA-U-TMCA
1.24 0.055
1.202 0.055
1.017 0.018
1.012 0.017
(a)
0.4 0.6 0.8 1 1.2 1.4 1.6
0
0.5
1
1.5
2
M
F
E
M
/
M
a
p
p
r
o
a
c
h
e
s
M
FEM
/ M
AA-Y-MMCA-UNI
M
FEM
/ M
AA-Y-WASA-UNI
M
FEM
/ M
NSA-U-WASA2
M
FEM
/ M
NSA-U-TMCA
M
FEM
/ M
approaches
mean c.o.v
1.35 0.092
1.234 0.041
1.054 0.043
1.046 0.045
(b)
Figure 4.15 Parametric study results
Series 1 Series 2
M
TEST
/M
approaches
y cr
= F F
y cr
= F F
96
Figure 4.15 (Continued)
0.2 0.4 0.6 0.8 1 1.2
0
0.5
1
1.5
2
M
F
E
M
/
M
a
p
p
r
o
a
c
h
e
s
M
FEM
/ M
AA-Y-MMCA
M
FEM
/ M
AA-Y-WASA
M
FEM
/ M
NSA-U-WASA2
M
FEM
/ M
NSA-U-TMCA
M
FEM
/ M
approaches
mean c.o.v
1.175 0.103
1.077 0.095
0.976 0.039
0.979 0.039
(c)
0.2 0.25 0.3 0.35 0.4 0.45 0.5 0.55 0.6 0.65
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
M
F
E
M
/
M
a
p
p
r
o
a
c
h
e
s
M
FEM
/ M
AA-Y-MMCA
M
FEM
/ M
NSA-U-TMCA
M
FEM
/ M
NSA-U-TMCA2
M
FEM
/ M
NSA-U-EPMC
M
FEM
/ M
NSA-U-ROMC
M
FEM
/ M
approaches
mean c.o.v
1.431 0.149
1.174 0.095
1.002 0.053
1.072 0.045
1.001 0.024
(d)
Series 1 Series 2
y
NA
=large y
NA
=small y
NA
=large y
NA
=small
y cr
= F F
y cr
= F F
97
Figure 4.15 (Continued)
0.2 0.4 0.6 0.8 1 1.2 1.4 1.6
0
0.5
1
1.5
2
2.5
M
F
E
M
/
M
a
p
p
r
o
a
c
h
e
s
M
FEM
/ M
AA-Y-MMCA
M
FEM
/ M
NSA-U-TMCA
M
FEM
/ M
NSA-U-TMCA2
M
FEM
/ M
approaches
mean c.o.v
1.649 0.207
1.119 0.08
1.065 0.07
(e)
0.2 0.4 0.6 0.8 1 1.2 1.4 1.6
0
0.5
1
1.5
2
2.5
M
F
E
M
/
M
a
p
p
r
o
a
c
h
e
s
M
FEM
/ M
AA-Y-MMCA
M
FEM
/ M
NSA-Y-TMCA
Series-1 Series-2 M
FEM
/ M
approaches
mean c.o.v
1.649 0.207
1.177 0.098
M
FEM
/ M
AA-Y-MMCA
M
FEM
/ M
NSA-Y-TMCA
Series-1 Series-2 M
FEM
/ M
approaches
mean c.o.v
1.649 0.207
1.177 0.098
Figure 4.16 Effect of idealization for dome-strut sections
y cr
= F F
y cr
= F F
98
4.6 Flexural Tests
The average of the three bending test results for an Aluminum Association
Standard I-Beam with uniform thickness (I-3x1.64), described in Section 2.8, is
included in Figure 4.15a within the dashed-line oval. The average of two other
bending test results for an American Standard I-Beam with tapered thickness (I-
3x1.96), listed in AA (2000b), is compared to the current and proposed approaches in
Table 4.5. From the comparisons, it can be seen that the proposed approaches are in
better agreement with the test results than the current approaches. Further details
regarding the tests are given in Section 3.6.
Table 4.5 Comparison of test results to proposed approaches for I-section with
tapered thickness (I-3x1.96)
current approaches proposed approaches
TESTS
AA-UNI-Y-MMCA
M
M
TESTS
AA-UNI-Y-WASA
M
M
TESTS
NSA-U-WASA2
M
M
TESTS
NSA-U-TMCA
M
M
1.174 1.141 0.967 0.960
Additionally, flexural tests are conducted for a complex extrusion type cross-
section. This cross-section is not a standard section but a mullion section made by
Kawneer Co., which is similar to the cross-section shown in Figure 4.7. The exact
dimensions of the cross-section cannot be presented here due to patent issues.
Before bending tests are conducted, the test setup is simulated using the finite
element method. The same solid elements are used to take into account the complex
geometry of the mullion section as those used for tapered thickness I-sections in
Section 3.6. On the assumption that the failure shape would be symmetric, only half of
the member is analyzed with longitudinal restraints at the cross-section cut. From the
analyses, the lateral support spacing is determined as shown in Figure 4.17 so that
continuous lateral support can be properly simulated. The test setup is designed for
typical two-point bending tests. Except for the spacing of the lateral supports specified
99
in this figure, details of the test setup are quite similar to those for the uniform
thickness I-sections shown in Figure 2.16.
Figure 4.17 Schematic bending test setup for mullion section
The residual deformations of the tested mullion sections are shown in Figure
4.18. In the first test, a single wave was formed near the middle of the span; this is
very similar to the assumption made for the finite element analysis that the failure
shape would be symmetric. On the other hand, in the second test, two waves were
formed; one is at the middle and the other is between two neighboring lateral supports.
(a)
(b)
(c)
(d)
Figure 4.18 Residual deformation of tested mullion specimens (a) side view at
Test 1 (b) plan view at Test 1 (c) side view at Test 2 (d) plan view at Test 2
100
Tests results are plotted and compared to the finite element analysis in Figure
4.19. Although there are some differences, the overall results are in good agreement.
The moment capacities from the tests and finite element simulation are compared to
those from the specification-based and proposed approaches in Table 4.6 and Table
4.7, respectively. From the tables, it is clear that the proposed approaches are in better
agreement with both the finite element analysis and the experimental test results than
current specification-based approaches for the mullion section.
0
0.2
0.4
0.6
0.8
1
1.2
1.4
0 10 20 30 40 50
Vertical Displacement (Span Center, mm)
L
o
a
d
F
a
c
t
o
r
(
M
u
/
M
y
)
Note. The maximum load factors are shown as solid shapes.
Figure 4.19 Test results for a mullion section
Table 4.6 Comparison of test results for mullion
current approaches proposed approaches
TESTS
AA-Y-MMCA
M
M
TESTS
NSA-U-TMCA
M
M
TESTS
NSA-U-TMCA2
M
M
1.295 1.096 1.106
Note. M
TESTS
is the average of the two maximum load factors from the tests.
Table 4.7 Comparison of finite element simulation results for mullion
current approaches proposed approaches
FEM
AA-Y-MMCA
M
M
FEM
NSA-U-TMCA
M
M
FEM
NSA-U-TMCA2
M
M
1.205 1.020 1.029
Max-LF Disp-XL
TEST 1 1.26 36.4
TEST 2 1.33 32.8
FEM 1.20 36.5
Note: Disp-XL = Displacement at max. LF
101
Before the tests are conducted, initial geometric imperfections were measured
on the compression flange and web as shown in Figure 6.5 in the Appendix. The
maximum imperfection on the flange is somewhat larger than the standard flatness
tolerance in AA (2000c), while the maximum imperfection on the web is much less
than the tolerance.
Uniaxial tension tests have also been conducted using seven test coupons
obtained from a portion adjacent to the specimens used for the bending tests. In Table
6.27d in the Appendix, the results are compared to the standard values listed in AA
(2000b). The median of the seven test results is introduced into the finite element
analysis.
4.7 Application to the AA Specification
In Section 2.9, two procedures are proposed for the governing allowable stress
equations for symmetric sections. In these procedures, the proposed allowable stress
equations for component elements are based on the AA Specification (2000a). By
replacing the allowable stress equations for component elements with the NSA, the
governing allowable stress equations can employ the NSA. In other words, Table 2.10
is replaced by Table 4.8.
It should be noted that the width-to-thickness ratio is not used for the
slenderness in Table 4.8. Instead, the equivalent slenderness ratio (
p
) is used directly
to determine the allowable stress. The equivalent slenderness ratio (
p
) is a function of
the buckling stress (F
cr
), as shown in Equation (47). The buckling stress is the
minimum value that can be obtained from any numerical buckling analysis program.
Further details regarding the NSA can be found in Section 4.1. Although Table 4.8 is
based on the Allowable Stress Design, this table can also be applied to the Load and
Resistance Factor Design by replacing the reciprocal of each safety factor with the
corresponding resistance factor.
102
Table 4.8 Allowable stress equations of NSA for (a) tension component element
(b) compression component element and (c) shape factors
(a)
AA Section allowable stress
ay y ty y
F F n =
3.4.2
3.4.4
( )
au u ty t u
F F k n =
(b)
c
AA Section
allowable stress
p
S
1
limit
S
1
allowable stress
S
1
p
S
2
limit
S
2
allowable stress
S
2
p
y cy
ay
y
F
F
n
=
y cy
B F
D
3.4.15
3.4.16
3.4.16.2
3.4.16.3
3.4.18, 3.4.19
u cy
au
u
F
F
n
=
y
u
n
u cy
n
B F
D
( )
1
ay au
p
y
F F
B D
n
= =
1
k B
D
2
ay au
y p
F F
k BE
n
= =
(c)
c
AA Section
yield shape
factor
ultimate shape factor
3.4.2 1.0
y
=
u tu ty
F F =
3.4.4 1.3
y
=
a
3
14 13 0.5 1.25 0.2
tu NA NA tu
u
ty ty
F y y F
F h h F
| | | |
| || |
= + + +
| |
| |
| |
\ .\ .
\ . \ .
3.4.15, 3.4.16
3.4.16.2, 3.4.16.3
1.0
y
=
b
u
= F
tu
/F
cy
3.4.18 1.3
y
=
b
3
14 13 0.5 1.25 0.2
tu NA NA tu
u
cy cy
F y y F
F h h F
| | | |
| || |
= + + +
| |
| |
| |
\ .\ .
\ . \ .
Note: B, D, k
1
and k
2
are the factors provided in the AA Specification (2000a). These factors differ for
flanges and webs. F
au
= ultimate allowable stress. F
ay
= yield allowable stress. y
NA
= the distance
from the neutral axis to the mid-depth. See Equation (47) for
p
.
a. In the AA Specification (2000a),
u
= 1.42F
tu
/F
ty
.
b. Not available in the AA Specification (2000a).
c. The allowable stress equations corresponding to the AA Section numbers indicated here can
be replaced with the NSA. In the following sections, flanges are under uniform compression,
while webs are under bending in own plane.
AA 3.4.2: tension flange
AA 3.4.4: tension web
AA 3.4.15: compression flange with one edge supported and the other edge free
AA 3.4.16: compression flange with both edges supported
AA 3.4.16.2: compression flange with one edge supported and the other edge with stiffener
AA 3.4.16.3: compression flange with both edges supported and with an intermediate stiffener
AA 3.4.18: compression web with both edges supported
AA 3.4.19: compression web with both edges supported with a horizontal stiffener
103
For cross-sections with the neutral axis not at mid-depth, the ultimate shape
factor of rectangular web elements (
u
) varies according to the location of the neutral
axis as shown in Equation (54). The maximum shape factor is obtained from one of
two solutions of 0
w NA
y = , which is when y
NA
= 0.125h (y
NA
= distance between
the neutral axis and the mid-depth of the web element. h = depth of the web element).
Introducing this into Equation (54) results in the maximum ultimate shape factor of
rectangular web elements when the neutral axis is not at mid-depth (unsymmetric
sections):
( )
0.125
max 1.34 0.286
NA
tu
u u
y h
ty
F
F
=
= = + (63)
1 1.5 2 2.5 3 3.5 4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
2.2
mean stdev min max
n
y
~
1.5 0.103 1.32 1.8
n
y
= 1.65
~
n
y
= 1.32
~
n
y
= 1.00
~
k
t
= 1.00
k
t
= 1.10
k
t
= 1.25
0 100 200 300
0.6
0.8
1
1.2
1.4
1.6
1.8
2
2.2
(a) (b)
Figure 4.20 Safety factor on yield strength of the tensile allowable stress
(AA Section 3.4.4) for a plate under bending when Procedure II is used
Using Equation (63), the safety factors on yield strength based on Procedure II
are computed for the previously selected 986 alloy-temper combinations as is done in
Section 2.9 for symmetric sections. Compared to the safety factor on yield strength in
the AA Specification (n
y
=1.65), the average of the varying safety factor (
y
n =1.50) is
y
n
tu ty
F F numbers of alloy-temper
104
9.1% smaller for unsymmetric cross-sections as shown in Figure 4.20. The average is
also somewhat smaller than that for symmetric cross-sections in Figure 2.20, as a
result of using a larger shape factor in Equation (63). However, the safety factor on
ultimate strength (n
u
= 1.95 for building and similar type of structures) given in the
AA Specification (2000a) is maintained, since only a specified percentage of the
ultimate allowable stress is used, as explained in Section 2.9.
4.8 Conclusions
The Numerical Slenderness Approach (NSA) is developed in this study to
provide a design approach for complex extrusions, in which the simply-supported
boundary idealization between component elements is not applicable or appropriate.
Since this approach computes the limit state (or allowable) stresses of component
elements using the numerically determined buckling stress, practically all aluminum
extrusions can be evaluated. The framework of the limit state (or allowable) stress
equations used for the AA Specification (2000a) is maintained in the NSA, with only
the expression for the equivalent slenderness ratio being changed.
For all cross-sections used in the parametric study, the proposed approaches
based on the NSA show a performance better than or equal to the approaches in the
AA Specification (2000a). The NSA shows a superior performance especially for
dome strut sections, in which intermediate stiffeners and variations in thickness are
ignored in the specification-based approaches. It is found that the NSA can be used for
the type of cross-sections that are presently covered by the AA Specification (2000a).
In addition, a wide variety of extruded aluminum sections that are not covered in the
specification can also be analyzed using the NSA.
Due to the flexibility of the extrusion process, many aluminum cross-sections
are not symmetric with respect to the bending axis or have a neutral axis that is not at
mid-depth. For these types of cross-sections, the rigorously determined shape factor of
105
rectangular web elements developed for symmetric sections in Section 2.3 is not
appropriate. Therefore, a rigorously determined shape factor based on the neutral axis
location for unsymmetric sections is developed in this chapter. The rigorous
expression is greatly simplified through curve-fittings for practical design purposes.
The Total Moment Capacity Approach (TMCA) developed for symmetric
sections in Section 2.6 is further extended to more general cross-sections in this study.
For sections with the neutral axis not at the mid-depth, TMCA2 is suggested, in which
the actual non-linear stress distribution is approximated by three linear stress
distributions. When the ultimate limit state is employed, TMCA2 shows good
agreement with the finite element analysis for the unsymmetric sections.
Moment capacity evaluation approaches relying on elasto-plastic stress
distribution (EPMC) and Ramberg-Osgood stress distribution (ROMC) are also
developed for unsymmetric sections. Although these approaches are quite accurate,
they are not appropriate for practical design approaches due to their complicated
procedures.
The approaches developed in this chapter are verified through parametric
studies using finite element analyses. In addition, the approaches are further validated
through physical flexural tests. In these tests, lateral supports are provided so that
lateral buckling is prevented.
Based on the NSA, the final form of the allowable stress equations has been
provided. The two procedures for calculating the governing allowable stress (or
factored limit state stress) suggested for symmetric sections have also been used for
unsymmetric sections in this chapter.
The ultimate shape factor for rectangular web elements with a neutral axis not
at mid-depth and the TMCA2 can also be employed in the modified specification-
based approaches in Chapter 2.
106
5. CONCLUSIONS
Aluminum is a structural material with a thousand faces, which are
combinations of a wide variety of both geometric shapes and material properties. This
wide range of variety distinguishes aluminum from other metals such as steel. To
make use of all the advantages of aluminum as a structural material, the variety should
be properly recognized.
In the study presented here, it is found that physical tests could be replaced
with numerical tools. Using numerical tools, a wide variety of virtual experiments can
be conducted. Most of the current specification approaches are based on physical tests
and analytical studies and it may not be appropriate to replace the whole specification
with results from numerical tools. Instead, if at least portions that are not covered in
the specification can be filled with the results of some simple numerical tools,
significant benefits can be expected. This study shows that linear elastic buckling
analyses using computers can improve the current specification in the application to a
wide variety of cross-sections obtained through extrusion.
In addition, the use of a uniform safety factor applied only to the yield strength
provided in the specification could be overconservative for certain alloy-temper
combinations that have a great margin between the ultimate stress and yield stress. For
this reason, a new design approach is proposed so that a larger inelastic reserve
capacity is recognized for such materials. This approach should provide a way to
make the current specifications more flexible with respect to material variations with a
varying safety factor on yield strength, while maintaining the usual safety factor on
ultimate strength.
In Chapter 2, equations are developed for the compressive limit state stress at
the ultimate limit state. The rigorous closed-form expression derived for the ultimate
shape factor using analytic integration shows that the current shape factor available in
106
107
the AA Specification (2000a) is unconservative for materials with a large ultimate to
yield stress ratio. Thus, a better shape factor expression, simplified for practical design
purposes, has been proposed. In addition, the parametric study for component
elements shows that the existing limit state stress equations for the inelastic buckling
range can be extended linearly up to the shape factor times the yield stress. A
parametric study for I-shaped sections using the proposed approaches shows good
agreement with the finite element analysis. The study is also further supported by the
flexural tests of a standard section.
For practical design purposes, two procedures are suggested for specifications.
In Procedure I, the safety factor on yield strength used in the current AA Specification
(2000a) is maintained. In Procedure II, the safety factor is applied to the additional
inelastic capacity. Procedure II is more reasonable in that the safety factor varies
depending on the margin between the ultimate and yield stresses.
In Chapter 3, initial and geometric stiffness matrices for plates with tapered
thickness are derived for a finite strip analysis program. The buckling coefficients
obtained by this program show good agreement with previous research results.
Expressions for the plate buckling coefficients of tapered plates are proposed with
respect to the thickness variation ratio. Using the expressions, limit state stress
equations are suggested for tapered thickness elements within the framework of the
current specification. A parametric study validates the developed equations for tapered
thickness plate flanges. Physical test results also agree well with the proposed
approaches.
In Chapter 4, the Numerical Slenderness Approach (NSA) is developed in this
study to provide a design approach for complex extrusions, in which the simply-
supported boundary idealization between component elements is not applicable or
appropriate. Since in this approach the limit state (or allowable) stresses of component
elements are computed using the numerically determined buckling stress, a wider
108
range of geometric shapes of the cross-section can be covered. The framework of the
limit state (or allowable) stress equations used for the specification is maintained in
the NSA, with only the expression for the equivalent slenderness ratio being changed.
For all cross-sections used in the parametric study, the proposed approaches based on
the NSA show a performance better than or equal to the modified specification-based
approaches in Chapters 2 and 3. In addition, a rigorous shape factor based on the
neutral axis location is developed, and it is greatly simplified for practical design
purposes. The Total Moment Capacity Approach (TMCA) developed for symmetric
sections in Chapter 2 is further extended to unsymmetric sections in the TMCA2,
where the actual non-linear stress distribution is approximated by three linear stress
distributions. Moment capacity evaluation approaches relying on elasto-plastic stress
distribution (EPMC) and Ramberg-Osgood stress distribution (ROMC) are also
developed for unsymmetric sections. Although these approaches are quite accurate,
they are not practical for use in design due to complicated procedures. The approaches
developed in this chapter are verified through parametric studies using finite element
analyses and physical flexural tests.
109
6. APPENDIX
6.1 Computational Example (1)
The following examples provide step-by-step computations of the moment
capacities based on the currently available two approaches in the AA Specification
(2000a) and two procedures proposed in Table 2.9. The safety factors on yield
strength corresponding to the moment capacities are also computed. The cross
sectional geometry is given in Figure 6.1, which is one of the cross-sections used in
the parametric study in Table 2.8.
Note: All dimensions are in inches and not to scale
Figure 6.1 Geometry of cross-section for Computational Example (1)
Given:
Cross section: Figure 6.1
Alloy-temper: 6061-T6. See the first alloy-temper combination in Table 2.4.
Required:
Allowable moment capacities and safety factors on yield strength
Solution:
In Table 6.1, sectional properties are calculated for the cross-section given in
Figure 6.1. For the computations of the ultimate shape factor for web (
u
), the
expression in the AA Specification, Equation (15), should be used, if the moment
capacity computation is based on the AA Specification. Similarly, the proposed shape
109
110
factor expression, Equation (16), should be used, if the moment capacity is based on
the proposed approaches. However, such consistency is not maintained in this
example, because the shape factors computed from the two equations do not differ
significantly for the material used here. Thus, Equation (16) is only used in this
example. The following computations are based on the sequence of table numbers
(from Table 6.1 to Table 6.6).
Table 6.1 Geometrical and material properties
geometrical properties material properties
I
f
= (10)(10.625
3
9.375
3
)/12 = 312.91 in
4
I
w
= (0.5)(9.375
3
)/12 = 34.33 in
4
I = I
f
+ I
w
= 347.24 in
4
S = I
/(h/2) = 65.36 in
3
S
f
= I
f
/(h
c
/2) = 62.58 in
3
S
w
= I
w
/(h
o
/2) = 7.324 in
3
A
f
=(10)(0.625) = 6.25 in
2
A
w
=(0.5)(9.375)/2 = 2.344 in
2
b = (10 0.5)/2 = 4.75 in
b/t
f
= 4.75/0.625 = 7.6
h
o
/t
w
= 9.375/0.5 = 18.75
F
tu
= 38 ksi
F
ty
= 35 ksi
F
cy
= 35 ksi
E = 10100 ksi (for compression)
Table 6.2 Buckling constants, shape factors, and slenderness limits (from Table
2.10)
flange web
B
p
= 45.04
D
p
= 0.3008
k
1
= 0.35
k
2
= 2.27
u
= 38/35 = 1.0857
ultimate limit state
( )( )
( )
1.65
1.95
1
45.04 1.0857 35
8.40
5.1 0.3008
u
S
= =
yield limit state
( )
1
45.04 35
6.54
5.1 0.3008
y
S
= =
( )
( )
2
0.35 45.04
10.28
5.1 0.3008
S = =
B
br
= 66.76
D
br
= 0.6648
k
1
= 0.5
k
2
= 2.04
u
= 1.25(38/35) + 0.2 = 1.557
ultimate limit state
( )( )
( )
1.65
1.95
1
66.76 1.557 35
46.36
0.67 0.6648
u
S
= =
yield limit state
( )
( )
1
66.76 1.3 35
47.73
0.67 0.6648
y
S
= =
( )
( )
2
0.5 66.76
74.94
0.67 0.6648
S = =
Note: B
p
, D
p
, B
br
, D
br
, k
1
, and k
2
are from AA (2000a)
111
Table 6.3 Allowable stresses for component elements (from Table 2.10)
flange web
For tension side,
35
21.21
1.65
ay
F ksi = =
( )( ) 1.0857 35
19.49
1.95
au
F ksi = =
For compression side,
since S
1
y
< b/t
f
< S
2
,
( )( )( ) 45.04 0.3008 5.1 7.6
20.23
1.65
ay
F ksi
= =
since b/t
f
< S
1
u
< S
2
, (not in AA)
( )( ) 1.0857 35
19.49
1.95
au
F ksi = =
For tension side,
( )( ) 1.3 35
27.58
1.65
ay
F ksi = =
( )( ) 1.557 35
27.95
1.95
au
F ksi = =
For compression side,
since h
o
/t
w
< S
1
y
< S
2
,
( )( ) 1.3 35
27.58
1.65
ay
F ksi = =
since h
o
/t
w
< S
1
u
< S
2
, (not in AA)
( )( ) 1.557 35
27.95
1.95
au
F ksi = =
Table 6.4 Member and governing allowable stresses (Table 2.7 and Table 2.9)
approaches member allowable stress
AA-MMCA F
a
= min (19.49, 21.21, 20.23, 27.95, 27.58, 27.58) = 19.49 ksi
AA-WASA
Tension
( ) ( ) 21.21 6.25 27.58 2.344 / 3
21.92
6.25 2.344/ 3
ay
F
+
= =
+
ksi
( ) ( )
( )
19.49 6.25 27.95 2.344 / 3
20.43
6.25 2.344 / 3
au
F
+
= =
+
ksi
Compression
( ) ( )
( )
20.23 6.25 27.58 2.344 / 3
21.05
6.25 2.344 / 3
ay
F
+
= =
+
ksi
Entire section
F
a
= min (21.92, 20.43, 21.05) = 20.43
Procedure I -
WASA2
Tension
( ) 21.92 10.625 10
ay
F = = 23.29 ksi
( ) 20.43 10.625 10
au
F = = 21.71 ksi
Compression
( ) 21.05 10.625 10
ay
F = =22.37 ksi
( ) ( )
( )
19.49 6.25 27.95 2.344 / 3
10.625
6.25 2.344 / 3 10
au
F
+
| |
= =
|
+
\ .
21.71 ksi
Entire section
F
a
= min (23.29, 21.71, 22.37, 21.71) = 21.71 ksi
112
Table 6.4 (Continued)
Procedure I -
TMCA
Tension
( ) ( ) 21.21 62.58 27.58 7.324
65.36
ay
F
+
= =23.40 ksi
( ) ( ) 19.49 62.58 27.95 7.324
65.36
au
F
+
= =21.79 ksi
Compression
( ) ( ) 20.23 62.58 27.58 7.324
65.36
ay
F
+
= = 22.46 ksi
( ) ( ) 19.49 62.58 27.95 7.324
65.36
au
F
+
= =21.79 ksi
Entire section
F
a
= min (23.40, 21.79, 22.46, 21.79) = 21.79 ksi
Procedure II -
WASA2
Tension
F
a
= 23.29+0.25(21.71
23.29)
= 22.90 < min{1.25(23.29), 21.71}?
No! F
a
= 21.71 ksi (due to the second safeguard)
Compression
F
a
= 21.71 ksi (due to the second safeguard)
Entire section
F
a
= 21.71 ksi
Procedure II -
TMCA
Tension
F
a
= 21.79 ksi (due to the second safeguard)
Compression
F
a
= 21.79 ksi (due to the second safeguard)
Entire section
F
a
= 21.79 ksi
Table 6.5 Moment capacity based on the yield limit state stresses and TMCA
yield limit
state stresses-
TMCA
F
LS
= F
ay
n
y
Tension
F
LSt
= 23.40(1.65) = 38.61 ksi
Compression
F
LSc
= 22.46(1.65) = 37.06 ksi
Entire section
F
LS
= min (38.61, 37.06) = 37.06 ksi
M
LS
= 37.06(65.36) = 2422.2 in-k
Note: To obtain F
LS
, F
ay
in Table 6.4 is multiplied by the yield safety factor (1.65).
113
Table 6.6 Allowable moment capacity and varying safety factor on yield strength
approaches
allowable moment capacity
( )
a a
M F S =
safety factor on yield strength
( )
y LS a
n M M =
AA-MMCA 19.49(65.36) = 1273.9 in-k 2422.2/1273.9 = 1.90 current AA
AA-WASA 20.43(65.36) = 1335.3 in-k 2422.2/1335.3 = 1.81
P1-WASA2 21.71(65.36) = 1419.0 in-k 2422.2/1419.0 = 1.71
P1-TMCA 21.79(65.36) = 1424.2 in-k 2422.2/1424.2 = 1.70
P2-WASA2 21.71(65.36) = 1419.0 in-k 2422.2/1419.0 = 1.71
proposed
procedures
P2-TMCA 21.79(65.36) = 1424.2 in-k 2422.2/1424.2 = 1.70
Note: In actual design computations, it is not necessary to compute the safety factors. However,
the safety factors are computed here just for comparison purposes.
6.2 Computational Example (2)
Given:
Cross section: Figure 6.1
Alloy-temper: 6063-T5 (test coupon diameter or thickness from 0.5 to 1.0 in.).
See the third alloy-temper combination in Table 2.4.
Required:
Allowable moment capacities and safety factors on yield strength
Solution:
All comments are the same as Computational Example (1). The following
computations are based on the sequence of table numbers (Table 6.7 to Table 6.12).
Table 6.7 Geometrical and material properties
geometrical properties material properties
I
f
= (10)(10.625
3
9.375
3
)/12 = 312.91 in
4
I
w
= (0.5)(9.375
3
)/12 = 34.33 in
4
I = I
f
+ I
w
= 347.24 in
4
S = I
/(h/2) = 65.36 in
3
S
f
= I
f
/(h
c
/2) = 62.58 in
3
S
w
= I
w
/(h
o
/2) = 7.324 in
3
A
f
=(10)(0.625) = 6.25 in
2
A
w
=(0.5)(9.375)/2 = 2.344 in
2
b = (10 0.5)/2 = 4.75 in
b/t
f
= 4.75/0.625 = 7.6
h
o
/t
w
= 9.375/0.5 = 18.75
F
tu
= 15 ksi
F
ty
= 21ksi
F
cy
= 15 ksi
E = 10100 ksi (for compression)
114
Table 6.8 Buckling constants, shape factors, and slenderness limits (from Table
2.10)
flange web
B
p
= 18.25
D
p
= 0.0775
k
1
= 0.35
k
2
= 2.27
u
= 21/15 = 1.4
ultimate limit state
( )( )
( )
1.65
1.95
1
18.25 1.4 15
1.22
5.1 0.0775
u
S
= =
yield limit state
( )
1
18.25 15
8.22
5.1 0.0775
y
S
= =
( )
( )
2
0.35 18.25
16.16
5.1 0.0775
S = =
B
p
= 26.37
D
p
= 0.1650
k
1
= 0.5
k
2
= 2.04
u
= 1.25(21/15) + 0.2 = 1.95
ultimate limit state
( )( )
( )
1.65
1.95
1
26.37 1.95 15
14.65
0.67 0.1650
u
S
= =
yield limit state
( )
( )
1
26.37 1.3 15
62.14
0.67 0.1650
y
S
= =
( )
( )
2
0.5 26.37
119.27
0.67 0.1650
S = =
Table 6.9 Allowable stress for component elements (from Table 2.10)
flange web
For tension side,
15
9.09
1.65
ay
F ksi = =
( )( ) 1.4 15
10.77
1.95
au
F ksi = =
For compression side,
since b/t
f
< S
1
y
< S
2
,
15
9.09
1.65
ay
F ksi = =
since S
1
u
< b/t
f
< S
2
,
( )( )( ) 18.25 0.0775 5.1 7.6
1.65
au
F
=
= 9.24 ksi
For tension side,
( )( ) 1.3 15
11.82
1.65
ay
F ksi = =
( )( ) 1.95 15
15
1.95
au
F ksi = =
For compression side,
since h
o
/t
w
< S
1
y
< S
2
,
( )( ) 1.3 15
11.82
1.65
ay
F ksi = =
since S
1
u
< h
o
/t
w
< S
2
,
( )( )( ) 26.37 0.1650 0.67 18.75
1.65
au
F
=
= 14.73 ksi
115
Table 6.10 Member and governing allowable stresses (Table 2.7 and Table 2.10)
approaches member allowable stress
AA-MMCA F
a
= min (9.09, 10.77, 9.09, 11.82, 15, 11.82) = 9.09 ksi
AA-WASA
Tension
( ) ( )
( )
9.09 6.25 11.82 2.344 / 3
9.39
6.25 2.344 / 3
ay
F
+
= =
+
ksi
( ) ( )
( )
10.77 6.25 15 2.344 / 3
11.24
6.25 2.344 / 3
au
F
+
= =
+
ksi
Compression
( ) ( )
( )
9.09 6.25 11.82 2.344 / 3
9.39
6.25 2.344 / 3
ay
F
+
= =
+
ksi
Entire section
F
a
= min (9.39, 11.24, 9.39) = 9.39 ksi
Procedure I -
WASA2
Tension
( ) 9.39 10.625 10
ay
F = =9.98 ksi
( ) 11.24 10.625 10
au
F = = 11.94 ksi
Compression
( ) 9.39 10.625 10
ay
F = =9.98 ksi
( ) ( )
( )
9.24 6.25 14.73 2.344 / 3
10.625
6.25 2.344 / 3 10
au
F
+
| |
= =
|
+
\ .
10.47 ksi
Entire section
F
a
= min (9.98, 11.94, 9.98, 10.47) = 9.98 ksi
Procedure I -
TMCA
Tension
( ) ( ) 9.09 62.58 11.82 7.324
65.36
ay
F
+
= = 10.03 ksi
( ) ( ) 10.77 62.58 15 7.324
65.36
au
F
+
= =11.99 ksi
Compression
( ) ( ) 9.09 62.58 11.82 7.324
65.36
ay
F
+
= = 10.03 ksi
( ) ( ) 9.24 62.58 14.73 7.324
65.36
au
F
+
= = 10.49 ksi
Entire section
F
a
= min (10.03, 11.99, 10.03, 10.49) = 10.03 ksi
116
Table 6.10 (Continued)
Procedure II -
WASA2
Tension
F
a
= 9.98+0.25(11.94 9.98) = 10.47 < min{1.25(9.98), 11.94}
Compression
F
a
= 9.98+0.25(10.47 9.98) = 10.10 < min{1.25(9.98), 10.47}
Entire section
F
a
= min(10.47, 10.10) = 10.10 ksi
Procedure II -
TMCA
Tension
F
a
= 10.03+0.25(11.99 10.03) = 10.52 < min{1.25(10.03), 11.99}
Compression
F
a
= 10.03+0.25(10.49 10.03) = 10.15 < min{1.25(10.03), 10.49}
Entire section
F
a
= min(10.52, 10.15) = 10.15 ksi
Table 6.11 Moment capacity based on the yield limit state stresses and TMCA
yield limit
state stresses-
TMCA
F
LS
= F
ay
n
y
Tension
F
LSt
= 10.03(1.65) = 16.55 ksi
Compression
F
LSc
= 10.03(1.65) = 16.55 ksi
Entire section
F
LS
= min (16.55, 16.55) = 16.55 ksi
M
LS
= 16.55(65.36) = 1081.7 in-k
Note: F
ay
from [Procedure I TMCA] in Table 6.10 is multiplied by the safety factor on yield strength.
Table 6.12 Allowable moment capacity and varying safety factor on yield
strength
approaches
allowable moment capacity
( )
a a
M F S =
safety factor on yield strength
( )
y LS a
n M M =
AA-MMCA 9.09(65.36) = 594.1 in-k 1081.7/594.1 = 1.82 current AA
AA-WASA 9.39(65.36) = 613.7 in-k 1081.7/613.7 = 1.76
P1-WASA2 9.98(65.36) = 652.3 in-k 1081.7/652.3 = 1.66
P1-TMCA 10.03(65.36) = 655.6 in-k 1081.7/655.6 = 1.65
P2-WASA2 10.10(65.36) = 660.1 in-k 1081.7/660.1 = 1.64
proposed
procedures
P2-TMCA 10.15(65.36) = 663.4 in-k 1081.7/663.4 = 1.63
Note: In actual design computations, it is not necessary to compute the safety factors. However,
the safety factors are computed here just for comparison purposes.
117
6.3 Computational Example (3)
The following examples provide step-by-step computations of the moment
capacities using the approaches based on the AA Specification (2000a) and the
Numerical Slenderness Approach. The results from these approaches without safety
factors are compared to the finite element analysis results. The cross sectional
geometry is given in Figure 6.2, which is one of the cross-sections used in the
parametric study in Chapter 4.
Note: All dimensions are in inches and not to scale
Figure 6.2 Geometry of cross section for Example (3)
Given:
Cross section: Figure 6.2
Alloy-temper: 6061-T6. See the first alloy-temper combination in Table 2.4.
Required:
Allowable moment capacities
Solution:
In Table 6.13, the sectional properties are calculated for the cross-section
given in Figure 6.2. The following computations are based on the sequence of table
118
numbers: from Table 6.13 to Table 6.16 for the AA Specification approach and from
Table 6.17 to Table 6.20 for the proposed approach.
Table 6.13 Geometrical and material properties
geometrical properties material properties
I
cf
= 4.390 in
4
(all compression flange group)
I
tf
= 4.932 in
4
(all tension flange group)
I
f
= 9.322 in
4
(all flange group)
I
w
= 1.628 in
4
(all web group)
I = 10.950 in
4
(all)
S = (10.950)/(2.5 + 0.309 + 0.125/2) =3.813 in
3
M
y
= F
y
S = (35)(3.813) = 133.47 in-k
b = (5 0.2343)/2 = 2.383 in
b/t
f
= 2.383/0.125 = 19.1
h
o
= 5 0.125 = 4.875 in
h
o
/t
w
= 4.875/0.0781 = 62.4
y
NA
= 0.309
F
tu
= 38 ksi
F
ty
= 35 ksi
F
cy
= 35 ksi
E = 10100 ksi (for compression)
Table 6.14 Buckling constants, shape factors, and slenderness limits for AA (from
Table 2.10)
flange web
B
p
= 45.04
D
p
= 0.3008
k
1
= 0.35
k
2
= 2.27
u
= 38/35 = 1.0857
ultimate limit state
( )( )
( )
1.65
1.95
1
45.04 1.0857 35
8.40
5.1 0.3008
u
S
= =
yield limit state
( )
1
45.04 35
6.54
5.1 0.3008
y
S
= =
( )
( )
2
0.35 45.04
10.28
5.1 0.3008
S = =
B
br
= 66.76
D
br
= 0.6648
k
1
= 0.5
k
2
= 2.04
a = 14x38/35+13 = 28.2: from Equation (52)
m = 3: from Equation (53)
y
NA
/h = 0.309/(5-0.125) = 0.0634
wo
= 1.25x38/35 + 0.2 = 1.557
u
= (28.2)(0.0634)(0.5-0.0634)
3
+ 1.557
= 1.706
ultimate limit state
( )( )
( )
1.65
1.95
1
66.76 1.706 35
36.45
0.67 0.6648
u
S
= =
yield limit state
( )
( )
1
66.76 1.3 35
47.73
0.67 0.6648
y
S
= =
( )
( )
2
0.5 66.76
74.94
0.67 0.6648
S = =
Note: B
p
, D
p
, B
br
, D
br
, k
1
, and k
2
are obtained from AA (2000a)
119
Table 6.15 Allowable stresses for component elements (from Table 2.10)
flange web
For tension side,
35
21.21
1.65
ay
F ksi = =
( )( ) 1.0857 35
19.49
1.95
au
F ksi = =
For compression side,
since b/t
f
> S
2
,
F
au
=
2.27 (45.04)(10100)
1.65(5.1)(19.06)
ay
F =
= 9.55 ksi
For tension side,
( )( ) 1.3 35
27.58
1.65
ay
F ksi = =
( )( ) 1.706 35
30.62
1.95
au
F ksi = =
For compression side,
since S
1
u
<S
1
y
< h
o
/t
w
< S
2
,
F
au
=
( )( )( ) 66.76 0.6648 0.67 62.4
1.65
ay
F
=
=23.62 ksi
Table 6.16 Member and governing allowable moments (from Table 2.7 and Table
2.9)
approaches member allowable moment
AA-MMCA
tension flange
M
ay
=
( ) 21.21 10.950
2.5 0.309 0.125/ 2 + +
= 80.88 in-k
M
au
=
( ) 19.49 10.950
2.5 0.309 0.125/ 2 + +
= 74.32 in-k
compression flange
M
ay
=
( ) 9.55 10.950
2.5 0.309
= 47.73 in-k
compression stiffeners (it is assumed that they reach the maximum)
M
ay
=
( ) 21.21 10.950
2.5 0.309 0.125/ 2 0.5 + +
= 84.35 in-k
tension web
M
ay
=
( ) 27.58 10.950
2.5 0.309 0.125/ 2 +
= 110.0 in-k
M
au
=
( ) 30.62 10.950
2.5 0.309 0.125/ 2 +
= 122.1 in-k
compression web
M
ay
=
( ) 23.62 10.950
2.5 0.309 0.125/ 2
= 121.5 in-k
M
a
= min(80.88, 74.32, 47.73, 84.35, 110.0, 122.1, 121.5)
= 47.73 in-k
AA-WASA Not appropriate
120
Table 6.17 Buckling constants, shape factors, and slenderness limits for NSA
(from Table 4.8)
Flange group Web group
B
p
, D
p
, k
1
, k
2
,
u
are the same as those
for AA in Table 6.14.
ultimate limit state
( )( )
1.65
1.95
1
45.04 1.0857 35
42.84
0.3008
u
S
= =
yield limit state
( )
1
45.04 35
33.38
0.3008
y
S
= =
( )
( )
2
0.35 45.04
52.41
0.3008
S = =
B
br
, D
br
, k
1
, k
2
,
u
are the same as those
for AA in Table 6.14.
ultimate limit state
( )( )
( )
1.65
1.95
1
66.76 1.706 35
24.42
0.6648
u
S
= =
yield limit state
( )
( )
1
66.76 1.3 35
31.98
0.6648
y
S
= =
( )
( )
2
0.5 66.76
50.21
0.6648
S = =
Note: B
p
, D
p
, B
br
, D
br
, k
1
, and k
2
are obtained from AA (2000a)
0
1
2
3
1 10 100
Figure 6.3 CUFSM analysis results finding minimum local bukling stress
From the CUFSM analysis results shown in Figure 6.3, the minimum buckling stress
is F
cr
= 1.02F
y
= 35.7 ksi. Thus, from Equation (47),
p cr
E F = = 10100 35.7
= 52.8. This is the equivalent slenderness ratio used in the NSA, which is the same for
all elements.
(11, 1.02)
M
c
r
/
M
y
half-wavelength (in.)
121
Table 6.18 Allowable stresses for component elements (from Table 4.8)
flange group web group
For tension side,
35
21.21
1.65
ay
F ksi = =
( )( ) 1.0857 35
19.49
1.95
au
F ksi = =
For compression side,
since
p
> S
2
, F
au
=
2.27 (45.04)(10100)
17.57
1.65(52.8)
ay
F ksi = =
For tension side,
( )( ) 1.3 35
27.58
1.65
ay
F ksi = =
( )( ) 1.706 35
30.62
1.95
au
F ksi = =
For compression side,
since
p
> S
2
, F
au
=
2.04 (66.76)(10100)
19.20
1.65(52.8)
ay
F ksi = =
Table 6.19 Member and governing allowable moments (Table 2.7 and Table 2.9)
NSA-TMCA-
Procedure I
Flange group
tension flange
M
ay
=
( ) 21.21 9.322
2.5 0.309 0.125/ 2 + +
= 68.86 in-k
M
au
=
( ) 19.49 9.322
2.5 0.309 0.125/ 2 + +
= 63.27 in-k
compression flange
M
ay
= M
au
=
( ) 17.57 9.322
2.5 0.309
= 74.75 in-k
compression stiffeners (it is assumed that they reach the maximum)
M
ay
=
( ) 21.21 9.322
2.5 0.309 0.125/ 2 0.5 + +
= 71.81 in-k
M
au
=
( ) 19.49 9.322
2.5 0.309 0.125/ 2 0.5 + +
= 65.98 in-k
M
f
= min(68.86, 63.27, 74.75, 71.81, 65.98) = 63.27 in-k
Web group
tension web
M
ay
=
( ) 27.58 1.628
2.5 0.309 0.125/ 2 +
= 16.35 in-k
M
au
=
( ) 30.62 1.628
2.5 0.309 0.125/ 2 +
= 18.15 in-k
compression web
M
ay
= M
au
=
( ) 19.20 1.628
2.5 0.309 0.125/ 2
= 14.69 in-k
M
w
= min(16.35, 18.15, 14.69) = 14.69 in-k
Entire section
M
a
= M
f
+ M
w
= 63.27 + 14.69 = 77.96 in-k
122
Table 6.19 (Continued)
NSA-TMCA-
Procedure II
Flange group
M
ay
= min(68.86, 74.75, 71.81) = 68.86 in-k
M
au
= min(63.27, 74.75, 65.98) = 63.27 in-k
M
f
= 63.27 in-k (due to a safeguard of Procedure II)
Web group
M
ay
= min(16.35, 14.69) = 14.69 in-k
M
au
= min(18.15, 14.69) = 14.69 in-k
M
w
= 14.69 in-k
Entire section
M
a
= M
f
+ M
w
= 63.27 + 14.69 = 77.96 ksi (same as Procedure I)
NSA-TMCA2-
Procedure I
All computations should be done without yield safety factor
(1.65).
Stress values computed here are based on the yield limit state
Flange group
F
cf
= 1.65(17.57) = 28.99 ksi
From Ramberg-Osgood equation with
u
= 8% assumption
cf
= 28.99/10100+0.002(28.99/35)
44.27
= 0.00287
From similar triangles,
tf
= 0.00287(2.5+0.309+0.125/2)/(2.5-0.309)=0.00376
By monotonically increasing trial values of stress using Ramberg-
Osgood equation, 0.00376 33.8/10100+0.002(33.8/35)
44.27
F
tf
= 33.8 ksi
Web group
F
cw
= (19.2)1.65 = 31.68 ksi
Element moment capacities
M
cf
= (28.99)(4.390)/(2.5-0.309) = 58.09 in-k
M
tf
= (33.8)(4.932)/(2.5+0.309+0.125/2) = 58.05 in-k
M
cw
= (31.68)(1.628)/(2.5-0.309-0.125/2) = 24.23 in-k
M
tw
= (27.58x1.65)(1.628)/(2.5+0.309-0.125/2) = 26.97 in-k
M
w
= min(24.23, 26.97) = 24.23 in-k
Entire section
M
u
= M
cf
+ M
tf
+ M
w
= 58.09 + 58.05 + 24.23 = 140.37 in-k
Table 6.20 Comparison of computed moment capacities
approaches
allowable moment
capacity (M
a
)
moment capacity without
safety factor (M
u
)
current AA AA-MMCA 47.73 in-k 47.73x1.65 = 78.75 in-k
P1-, P2-TMCA 77.96 in-k 77.96 x 1.65 = 128.63 in-k proposed
procedures
P1-TMCA2 140.37/1.65 = 85.07 in-k 140.37 in-k
FEM 1.06(133.47) = 141.5 in-k
123
6.4 Dimensions and Moment Capacities of Parametric Study Sections
The tables and figures in this section provide detailed dimensions and analysis
results of the cross-sections used in the parametric studies in this report.
(a) (b) (c)
(d) (e) (f)
Note: All dimensions are in inches and not to scale.
Figure 6.4 Cross-sections used in the parametric study
Table 6.21 I-sections with uniform thickness for Figure 4.15a (Figure 6.4a)
t
w
t
f
M
NSA-U-WASA2
M
y
M
NSA-U-TMCA
M
y
t
w
t
f
M
NSA-U-WASA2
M
y
M
NSA-U-TMCA
M
y
25.4 42.3 1.309 1.316 8.46 42.3 1.282 1.284
25.4 31.8 1.277 1.285 8.46 31.8 1.241 1.244
25.4 21.2 1.249 1.259 8.46 21.2 1.164 1.167
25.4 15.9 1.213 1.224 8.46 15.9 1.075 1.079
25.4 12.7 1.156 1.167 8.46 12.7 1.002 1.005
12.7 42.3 1.289 1.292 6.35 42.3 1.278 1.280
12.7 31.8 1.251 1.255 6.35 31.8 1.236 1.238
12.7 21.2 1.194 1.199 6.35 21.2 1.150 1.152
12.7 15.9 1.119 1.125 6.35 15.9 1.053 1.056
12.7 12.7 1.059 1.065 6.35 12.7 0.971 0.974
Note: The moment capacities obtained by M
AA-Y-MMCA
, M
AA-Y-WASA
, M
AA-U-WASA2
, and
M
AA-U-TMCA
are given in Table 2.8. All dimensions are in mm.
124
Table 6.22 I-sections with tapered thickness for Figure 3.10 (Figure 6.4b)
(a) Series 1
(t
2
-t
1
) M
UNI-Y-MMCA
M
UNI-Y-WASA
M
TAP-U-WASA2
M
TAP-U-TMCA
M
FEM
Desig-
nation
h w
t
f
(mean)
t
w
b
0
M
y
M
y
M
y
M
y
M
y
y
cr
F
=
F
WF2x1.43 2.0 2.00 0.232 0.188 1/11.4 1.000 1.027 1.280 1.289 1.286 0.293
WF2.5x1.8 2.5 2.00 0.247 0.250 1/7.00 1.000 1.040 1.282 1.296 1.323 0.259
WF4x4.76 4.0 4.00 0.370 0.313 1/11.3 1.000 1.029 1.249 1.257 1.249 0.364
WF5x6.49 5.0 5.00 0.415 0.313 1/13.6 1.000 1.027 1.203 1.209 1.217 0.422
WF6x7.85 6.0 5.93 0.451 0.250 1/15.6 1.000 1.021 1.153 1.157 1.176 0.496
WF6x8.30 6.0 6.00 0.451 0.313 1/15.6 1.000 1.026 1.162 1.167 1.184 0.478
WF6x9.18 6.0 6.13 0.451 0.438 1/15.6 1.000 1.034 1.178 1.185 1.208 0.446
WF8x11.2 8.0 7.94 0.458 0.313 1/18.9 0.922 0.956 1.063 1.066 1.121 0.629
WF8x11.8 8.0 8.00 0.458 0.375 1/18.9 0.922 0.961 1.073 1.077 1.131 0.606
WF8x13.0 8.0 8.13 0.458 0.500 1/18.9 0.922 0.972 1.091 1.096 1.143 0.567
I3x1.96 3.0 2.33 0.257 0.170 1/18.9 1.000 1.029 1.245 1.254 1.268 0.315
I3x2.25 3.0 2.41 0.257 0.251 1/6.00 1.000 1.039 1.265 1.277 1.303 0.297
I3x2.59 3.0 2.51 0.257 0.349 1/6.00 1.000 1.050 1.284 1.301 1.333 0.284
I4x2.64 4.0 2.66 0.289 0.190 1/6.00 1.000 1.034 1.235 1.245 1.271 0.325
I4x3.28 4.0 2.80 0.289 0.326 1/6.00 1.000 1.052 1.266 1.282 1.322 0.298
I5x3.43 5.0 3.00 0.323 0.210 1/6.00 1.000 1.038 1.231 1.240 1.250 0.333
I5x4.23 5.0 3.14 0.323 0.347 1/6.00 1.000 1.055 1.261 1.277 1.314 0.305
I5x5.10 5.0 3.28 0.323 0.494 1/6.00 1.000 1.070 1.287 1.308 1.344 0.290
I6x4.30 6.0 3.33 0.355 0.230 1/6.00 1.000 1.041 1.229 1.238 1.243 0.339
I6x5.10 6.0 4.44 0.355 0.343 1/6.00 1.000 1.044 1.237 1.249 1.250 0.386
I6x5.96 6.0 3.57 0.355 0.465 1/6.00 1.000 1.068 1.276 1.294 1.331 0.299
I7x5.27 7.0 3.66 0.389 0.250 1/6.00 1.000 1.044 1.227 1.237 1.235 0.344
I7x6.05 7.0 3.76 0.389 0.345 1/6.00 1.000 1.055 1.248 1.261 1.289 0.325
I7x6.92 7.0 3.86 0.389 0.450 1/6.00 1.000 1.066 1.267 1.284 1.332 0.306
I8x6.35 8.0 4.00 0.421 0.270 1/6.00 1.000 1.045 1.227 1.236 1.235 0.351
I8x7.96 8.0 4.17 0.421 0.441 1/6.00 1.000 1.065 1.260 1.275 1.326 0.316
I8x8.81 8.0 4.26 0.421 0.532 1/6.00 1.000 1.073 1.275 1.293 1.328 0.305
I9x7.51 9.0 4.33 0.453 0.290 1/6.00 1.000 1.047 1.226 1.236 1.211 0.356
I10x8.76 10 4.66 0.487 0.310 1/6.00 1.000 1.048 1.225 1.235 1.228 0.360
I10x10.4 10 4.80 0.487 0.447 1/6.00 1.000 1.063 1.251 1.265 1.293 0.332
I10x12.1 10 4.94 0.487 0.594 1/6.00 1.000 1.076 1.274 1.292 1.328 0.309
I12x11.0 12 5.00 0.538 0.350 1/6.00 1.000 1.055 1.225 1.235 1.217 0.364
I12x12.1 12 5.08 0.538 0.428 1/6.00 1.000 1.063 1.245 1.258 1.236 0.345
I12x14.1 12 5.25 0.653 0.460 1/6.00 1.000 1.055 1.244 1.256 1.243 0.306
I12x15.6 12 5.36 0.653 0.565 1/6.00 1.000 1.064 1.259 1.273 1.301 0.291
I12x17.3 12 5.48 0.653 0.687 1/6.00 1.000 1.072 1.274 1.291 1.349 0.275
Note: 1. The designations given here represent the original ones in AA (2000b).
2. All dimensions are in inches.
125
Table 6.22 (Continued)
(b) Series 2 (60% thickness reduction)
(t
2
-t
1
) M
UNI-Y-MMCA
M
UNI-Y-WASA
M
TAP-U-WASA2
M
TAP-U-TMCA
M
FEM
Desig-
nation
h w
t
f
(mean)
t
w
b
0
M
y
M
y
M
y
M
y
M
y
y
cr
F
=
F
WF2x1.43 2.0 2.00 0.093 0.075 1/28.50 0.827 0.878 0.977 0.979 1.066 0.761
WF2.5x1.8 2.5 2.00 0.099 0.100 1/17.50 0.865 0.934 1.049 1.053 1.122 0.666
WF4x4.76 4.0 4.00 0.148 0.125 1/28.25 0.655 0.728 0.831 0.831 0.958 0.937
WF5x6.49 5.0 5.00 0.166 0.125 1/34.00 0.584 0.658 0.734 0.733 0.869 1.083
WF6x7.85 6.0 5.93 0.180 0.100 1/39.00 0.531 0.584 0.635 0.635 0.765 1.269
WF6x8.30 6.0 6.00 0.180 0.125 1/39.00 0.527 0.601 0.657 0.657 0.788 1.225
WF6x9.18 6.0 6.13 0.180 0.175 1/39.00 0.520 0.619 0.697 0.696 0.842 1.142
WF8x11.2 8.0 7.94 0.183 0.125 1/47.25 0.402 0.474 0.513 0.512 0.636 1.601
WF8x11.8 8.0 8.00 0.183 0.150 1/47.25 0.400 0.499 0.538 0.537 0.665 1.542
WF8x13.0 8.0 8.13 0.183 0.200 1/47.25 0.396 0.526 0.584 0.582 0.699 1.441
I3x1.96 3.0 2.33 0.103 0.068 1/15.00 0.780 0.838 0.965 0.968 1.036 0.807
I3x2.25 3.0 2.41 0.103 0.100 1/15.00 0.764 0.846 1.007 1.010 1.070 0.759
I3x2.59 3.0 2.51 0.103 0.140 1/15.00 0.744 0.852 1.029 1.034 1.086 0.721
I4x2.64 4.0 2.66 0.116 0.076 1/15.00 0.767 0.836 0.950 0.954 1.025 0.830
I4x3.28 4.0 2.80 0.116 0.130 1/15.00 0.743 0.852 1.022 1.026 1.075 0.756
I5x3.43 5.0 3.00 0.129 0.084 1/15.00 0.760 0.823 0.937 0.941 1.018 0.848
I5x4.23 5.0 3.14 0.129 0.139 1/15.00 0.739 0.854 1.011 1.016 1.068 0.773
I5x5.10 5.0 3.28 0.129 0.198 1/15.00 0.719 0.870 1.060 1.066 1.104 0.730
I6x4.30 6.0 3.33 0.142 0.092 1/15.00 0.752 0.810 0.924 0.928 1.004 0.865
I6x5.10 6.0 4.44 0.142 0.137 1/15.00 0.566 0.688 0.875 0.877 0.941 0.975
I6x5.96 6.0 3.57 0.142 0.186 1/15.00 0.720 0.866 1.043 1.048 1.088 0.754
I7x5.27 7.0 3.66 0.156 0.100 1/15.00 0.750 0.801 0.914 0.918 0.987 0.878
I7x6.05 7.0 3.76 0.156 0.138 1/15.00 0.737 0.851 0.967 0.971 1.038 0.825
I7x6.92 7.0 3.86 0.156 0.180 1/15.00 0.725 0.865 1.017 1.022 1.075 0.774
I8x6.35 8.0 4.00 0.168 0.108 1/15.00 0.742 0.788 0.903 0.907 0.988 0.895
I8x7.96 8.0 4.17 0.168 0.176 1/15.00 0.723 0.860 0.992 0.997 1.059 0.801
I8x8.81 8.0 4.26 0.168 0.213 1/15.00 0.714 0.871 1.031 1.036 1.088 0.768
I9x7.51 9.0 4.33 0.181 0.116 1/15.00 0.738 0.778 0.893 0.898 0.975 0.910
I10x8.76 10 4.66 0.195 0.124 1/15.00 0.737 0.772 0.886 0.891 0.965 0.921
I10x10.4 10 4.80 0.195 0.179 1/15.00 0.723 0.840 0.956 0.961 1.032 0.843
I10x12.1 10 4.94 0.195 0.238 1/15.00 0.711 0.873 1.018 1.023 1.080 0.780
I12x11.0 12 5.00 0.215 0.140 1/15.00 0.760 0.783 0.882 0.887 0.963 0.935
I12x12.1 12 5.08 0.215 0.171 1/15.00 0.752 0.822 0.922 0.928 1.007 0.881
I12x14.1 12 5.25 0.261 0.184 1/15.00 0.862 0.915 0.975 0.981 1.063 0.786
I12x15.6 12 5.36 0.261 0.226 1/15.00 0.856 0.952 1.013 1.019 1.102 0.743
I12x17.3 12 5.48 0.261 0.275 1/15.00 0.850 0.968 1.052 1.058 1.140 0.700
Note: 1. The designations given here are from AA (2000b). However, the thickness of all
component elements of Series 2 sections is reduced by 60% from the original ones.
2. All dimensions are in inches
126
Table 6.23 I-sections with tapered thickness for Figure 4.15b (Figure 6.4b)
(a) Series 1
(t
2
-t
1
) M
NSA-U-WASA2
M
NSA-U-TMCA
M
FEM
Desig-
nation
h w
t
f
(mean)
t
w
b
0
M
y
M
y
M
y
y
cr
F
=
F
WF2x1.43 2.0 2.00 0.232 0.188 1/11.4 1.280 1.289 1.286 0.293
WF2.5x1.8 2.5 2.00 0.247 0.250 1/7.00 1.282 1.296 1.323 0.259
WF4x4.76 4.0 4.00 0.370 0.313 1/11.3 1.249 1.258 1.249 0.364
WF5x6.49 5.0 5.00 0.415 0.313 1/13.6 1.226 1.234 1.217 0.422
WF6x7.85 6.0 5.93 0.451 0.250 1/15.6 1.174 1.180 1.176 0.496
WF6x8.30 6.0 6.00 0.451 0.313 1/15.6 1.190 1.197 1.184 0.478
WF6x9.18 6.0 6.13 0.451 0.438 1/15.6 1.219 1.228 1.208 0.446
WF8x11.2 8.0 7.94 0.458 0.313 1/18.9 1.087 1.092 1.121 0.629
WF8x11.8 8.0 8.00 0.458 0.375 1/18.9 1.104 1.110 1.131 0.606
WF8x13.0 8.0 8.13 0.458 0.500 1/18.9 1.134 1.143 1.143 0.567
I3x1.96 3.0 2.33 0.257 0.170 1/18.9 1.245 1.254 1.268 0.315
I3x2.25 3.0 2.41 0.257 0.251 1/6.00 1.265 1.277 1.303 0.297
I3x2.59 3.0 2.51 0.257 0.349 1/6.00 1.284 1.301 1.333 0.284
I4x2.64 4.0 2.66 0.289 0.190 1/6.00 1.235 1.245 1.271 0.325
I4x3.28 4.0 2.80 0.289 0.326 1/6.00 1.266 1.282 1.322 0.298
I5x3.43 5.0 3.00 0.323 0.210 1/6.00 1.231 1.240 1.250 0.333
I5x4.23 5.0 3.14 0.323 0.347 1/6.00 1.261 1.277 1.314 0.305
I5x5.10 5.0 3.28 0.323 0.494 1/6.00 1.287 1.308 1.344 0.290
I6x4.30 6.0 3.33 0.355 0.230 1/6.00 1.229 1.238 1.243 0.339
I6x5.10 6.0 4.44 0.355 0.343 1/6.00 1.231 1.243 1.250 0.386
I6x5.96 6.0 3.57 0.355 0.465 1/6.00 1.276 1.294 1.331 0.299
I7x5.27 7.0 3.66 0.389 0.250 1/6.00 1.227 1.237 1.235 0.344
I7x6.05 7.0 3.76 0.389 0.345 1/6.00 1.248 1.261 1.289 0.325
I7x6.92 7.0 3.86 0.389 0.450 1/6.00 1.267 1.284 1.332 0.306
I8x6.35 8.0 4.00 0.421 0.270 1/6.00 1.226 1.236 1.235 0.351
I8x7.96 8.0 4.17 0.421 0.441 1/6.00 1.260 1.275 1.326 0.316
I8x8.81 8.0 4.26 0.421 0.532 1/6.00 1.275 1.293 1.328 0.305
I9x7.51 9.0 4.33 0.453 0.290 1/6.00 1.224 1.234 1.211 0.356
I10x8.76 10 4.66 0.487 0.310 1/6.00 1.224 1.234 1.228 0.360
I10x10.4 10 4.80 0.487 0.447 1/6.00 1.251 1.265 1.293 0.332
I10x12.1 10 4.94 0.487 0.594 1/6.00 1.274 1.292 1.328 0.309
I12x11.0 12 5.00 0.538 0.350 1/6.00 1.227 1.238 1.217 0.364
I12x12.1 12 5.08 0.538 0.428 1/6.00 1.245 1.258 1.236 0.345
I12x14.1 12 5.25 0.653 0.460 1/6.00 1.244 1.256 1.243 0.306
I12x15.6 12 5.36 0.653 0.565 1/6.00 1.259 1.273 1.301 0.291
I12x17.3 12 5.48 0.653 0.687 1/6.00 1.274 1.291 1.349 0.275
Note: 1. The designations given here represent the original ones in AA
(2000b).
2. All dimensions are in inches.
127
Table 6.23 (Continued)
(b) Series 2 (60% thickness reduction)
(t
2
-t
1
) M
NSA-U-WASA2
M
NSA-U-TMCA
M
FEM
Desig-
nation
h w
t
f
(mean)
t
w
b
0
M
y
M
y
M
y
y
cr
F
=
F
WF2x1.43 2.0 2.00 0.093 0.075 1/28.50 1.007 1.012 1.066 0.761
WF2.5x1.8 2.5 2.00 0.099 0.100 1/17.50 1.065 1.073 1.122 0.666
WF4x4.76 4.0 4.00 0.148 0.125 1/28.25 0.903 0.907 0.958 0.937
WF5x6.49 5.0 5.00 0.166 0.125 1/34.00 0.791 0.794 0.869 1.083
WF6x7.85 6.0 5.93 0.180 0.100 1/39.00 0.671 0.673 0.765 1.269
WF6x8.30 6.0 6.00 0.180 0.125 1/39.00 0.697 0.699 0.788 1.225
WF6x9.18 6.0 6.13 0.180 0.175 1/39.00 0.750 0.753 0.842 1.142
WF8x11.2 8.0 7.94 0.183 0.125 1/47.25 0.529 0.531 0.636 1.601
WF8x11.8 8.0 8.00 0.183 0.150 1/47.25 0.550 0.552 0.665 1.542
WF8x13.0 8.0 8.13 0.183 0.200 1/47.25 0.590 0.593 0.699 1.441
I3x1.96 3.0 2.33 0.103 0.068 1/15.00 0.972 0.977 1.036 0.807
I3x2.25 3.0 2.41 0.103 0.100 1/15.00 1.006 1.013 1.070 0.759
I3x2.59 3.0 2.51 0.103 0.140 1/15.00 1.037 1.047 1.086 0.721
I4x2.64 4.0 2.66 0.116 0.076 1/15.00 0.957 0.962 1.025 0.830
I4x3.28 4.0 2.80 0.116 0.130 1/15.00 1.011 1.019 1.075 0.756
I5x3.43 5.0 3.00 0.129 0.084 1/15.00 0.945 0.950 1.018 0.848
I5x4.23 5.0 3.14 0.129 0.139 1/15.00 1.000 1.007 1.068 0.773
I5x5.10 5.0 3.28 0.129 0.198 1/15.00 1.037 1.047 1.104 0.730
I6x4.30 6.0 3.33 0.142 0.092 1/15.00 0.935 0.940 1.004 0.865
I6x5.10 6.0 4.44 0.142 0.137 1/15.00 0.876 0.882 0.941 0.975
I6x5.96 6.0 3.57 0.142 0.186 1/15.00 1.018 1.027 1.088 0.754
I7x5.27 7.0 3.66 0.156 0.100 1/15.00 0.927 0.932 0.987 0.878
I7x6.05 7.0 3.76 0.156 0.138 1/15.00 0.965 0.971 1.038 0.825
I7x6.92 7.0 3.86 0.156 0.180 1/15.00 1.002 1.010 1.075 0.774
I8x6.35 8.0 4.00 0.168 0.108 1/15.00 0.917 0.922 0.988 0.895
I8x7.96 8.0 4.17 0.168 0.176 1/15.00 0.983 0.991 1.059 0.801
I8x8.81 8.0 4.26 0.168 0.213 1/15.00 1.009 1.018 1.088 0.768
I9x7.51 9.0 4.33 0.181 0.116 1/15.00 0.909 0.914 0.975 0.910
I10x8.76 10 4.66 0.195 0.124 1/15.00 0.903 0.907 0.965 0.921
I10x10.4 10 4.80 0.195 0.179 1/15.00 0.955 0.962 1.032 0.843
I10x12.1 10 4.94 0.195 0.238 1/15.00 1.002 1.010 1.080 0.780
I12x11.0 12 5.00 0.215 0.140 1/15.00 0.895 0.900 0.963 0.935
I12x12.1 12 5.08 0.215 0.171 1/15.00 0.930 0.936 1.007 0.881
I12x14.1 12 5.25 0.261 0.184 1/15.00 0.986 0.992 1.063 0.786
I12x15.6 12 5.36 0.261 0.226 1/15.00 1.017 1.024 1.102 0.743
I12x17.3 12 5.48 0.261 0.275 1/15.00 1.051 1.059 1.140 0.700
Note: 1. The designations given here are from AA (2000b). However, the
thickness of all component elements of Series 2 sections is reduced by
60% from the original ones.
2. All dimensions are in inches
128
Table 6.24 Edge-stiffened Z-sections for Figure 4.15c (Figure 6.4c)
M
AA-Y-MMCA
M
AA-Y-WASA
M
NSA-U-WASA2
M
NSA-U-TMCA
M
FEM
t
f
t
w
t
s
d
s
M
y
M
y
M
y
M
y
M
y
y
cr
F
=
F
0.164 0.164 0.164 0.122 0.906 0.995 1.036 1.033 1.020 0.745
0.164 0.164 0.164 0.247 0.964 1.038 1.064 1.061 1.030 0.695
0.164 0.164 0.164 0.368 1.000 1.061 1.097 1.094 1.050 0.636
0.164 0.164 0.164 0.552 1.000 1.057 1.134 1.132 1.060 0.568
0.082 0.082 0.082 0.271 0.607 0.791 0.765 0.765 0.828 1.118
0.082 0.082 0.082 0.551 0.811 0.940 0.942 0.942 0.941 0.860
0.082 0.082 0.082 0.822 0.869 0.946 1.001 1.004 0.973 0.751
0.082 0.082 0.082 1.234 0.588 0.702 1.020 1.026 0.904 0.711
0.491 0.491 0.491 0.247 1.000 1.055 1.333 1.311 1.281 0.274
0.491 0.491 0.491 0.368 1.000 1.053 1.329 1.308 1.286 0.273
0.491 0.491 0.491 0.552 1.000 1.050 1.323 1.306 1.295 0.265
0.245 0.245 0.245 0.271 1.000 1.061 1.207 1.200 1.158 0.495
0.245 0.245 0.245 0.551 1.000 1.055 1.235 1.230 1.180 0.435
0.245 0.245 0.245 0.822 1.000 1.051 1.236 1.234 1.186 0.396
0.245 0.245 0.245 1.234 1.000 1.045 1.231 1.236 1.197 0.373
0.123 0.123 0.123 0.175 0.743 0.878 0.915 0.913 0.980 0.928
0.123 0.123 0.123 0.355 0.862 0.967 0.999 0.998 1.026 0.783
0.123 0.123 0.123 0.529 0.976 1.043 1.053 1.052 1.044 0.688
0.123 0.123 0.123 0.794 1.000 1.054 1.097 1.099 1.067 0.605
0.164 0.164 0.327 0.165 0.951 1.026 1.058 1.054 1.010 0.704
0.164 0.164 0.327 0.246 0.991 1.052 1.078 1.075 1.030 0.665
0.164 0.164 0.327 0.369 1.000 1.054 1.105 1.104 1.040 0.611
0.082 0.082 0.164 0.208 0.574 0.774 0.740 0.740 0.757 1.155
0.082 0.082 0.164 0.422 0.734 0.898 0.895 0.896 0.877 0.937
0.082 0.082 0.164 0.630 0.890 0.990 0.951 0.953 0.946 0.833
0.082 0.082 0.164 0.944 1.000 1.041 0.991 0.996 0.876 0.754
0.491 0.491 0.982 0.327 1.000 1.048 1.320 1.302 1.276 0.268
0.491 0.491 0.982 0.410 1.000 1.046 1.316 1.299 1.278 0.266
0.491 0.491 0.982 0.492 1.000 1.044 1.312 1.298 1.278 0.262
0.491 0.491 0.982 0.615 1.000 1.041 1.307 1.295 1.278 0.256
0.245 0.245 0.491 0.331 1.000 1.053 1.217 1.211 1.161 0.464
0.245 0.245 0.491 0.545 1.000 1.047 1.226 1.223 1.173 0.422
0.245 0.245 0.491 0.752 1.000 1.042 1.223 1.224 1.184 0.395
0.245 0.245 0.491 1.067 1.000 1.036 1.217 1.225 1.187 0.373
0.123 0.123 0.245 0.199 0.785 0.911 0.951 0.950 0.973 0.864
0.123 0.123 0.245 0.342 0.880 0.980 1.007 1.006 1.014 0.762
0.123 0.123 0.245 0.480 0.973 1.035 1.043 1.043 1.035 0.695
0.123 0.123 0.245 0.689 1.000 1.046 1.073 1.076 1.063 0.632
Note: All dimensions are in inches.
129
Table 6.25 Edge-stiffened unsymmetric sections for Figure 4.15d
(a) (Figure 6.4d)
M
AA-Y-MMCA
M
NSA-U-TMCA
M
NSA-U-TMCA2
M
NSA-U-EPMC
M
NSA-U-ROMC
M
FEM
w
t
y
NA
y
EP
y
RO
M
y
M
y
M
y
M
y
M
y
M
y
y
cr
F
=
F
17 -0.02 1.00 -0.07 1.007 1.155 1.194 1.189 1.277 1.264 0.299
16 0.10 1.50 0.32 1.000 1.152 1.211 1.200 1.291 1.286 0.293
15 0.23 2.00 0.70 1.000 1.157 1.238 1.220 1.313 1.317 0.285
14 0.36 2.50 1.07 1.000 1.163 1.267 1.238 1.336 1.347 0.277
13 0.50 2.78 1.43 1.000 1.169 1.297 1.255 1.360 1.382 0.269
12 0.65 2.97 1.78 1.000 1.176 1.330 1.272 1.385 1.411 0.261
11 0.80 3.13 2.10 1.000 1.184 1.365 1.291 1.410 1.454 0.252
10 0.96 3.26 2.41 1.000 1.193 1.403 1.312 1.438 1.489 0.243
9 1.13 3.38 2.70 1.000 1.203 1.444 1.336 1.468 1.501 0.234
8 1.31 3.50 2.96 1.000 1.215 1.489 1.365 1.502 1.544 0.224
7 1.50 3.60 3.20 1.000 1.229 1.540 1.398 1.541 1.592 0.214
6 1.70 3.70 3.42 1.000 1.247 1.598 1.438 1.588 1.648 0.203
5 1.91 3.80 3.62 1.000 1.269 1.664 1.488 1.645 1.696 0.192
4 2.14 3.90 3.79 1.000 1.297 1.744 1.550 1.718 1.756 0.180
3 2.38 4.00 3.95 1.000 1.336 1.842 1.632 1.815 1.846 0.167
(b) (Figure 6.4e)
M
AA-Y-MMCA
M
NSA-U-TMCA
M
NSA-U-TMCA2
M
NSA-U-EPMC
M
NSA-U-ROMC
M
FEM
w
t
y
NA
y
EP
y
RO
M
y
M
y
M
y
M
y
M
y
M
y
y
cr
F
=
F
16 0.09 0.17 0.25 1.000 1.069 1.044 1.027 1.055 1.078 0.668
15 0.21 0.63 0.61 1.000 1.120 1.081 1.058 1.095 1.109 0.651
14 0.34 1.07 0.98 1.000 1.131 1.121 1.087 1.135 1.139 0.634
13 0.48 2.78 1.35 1.000 1.145 1.165 1.188 1.174 1.168 0.615
12 0.62 2.97 1.80 1.000 1.161 1.214 1.206 1.215 1.200 0.598
11 0.78 3.13 2.28 1.000 1.175 1.263 1.224 1.258 1.230 0.578
10 0.93 3.26 2.66 1.000 1.183 1.308 1.245 1.299 1.262 0.559
9 1.10 3.38 2.95 1.000 1.193 1.357 1.269 1.341 1.296 0.539
8 1.27 3.50 3.19 1.000 1.204 1.411 1.296 1.386 1.334 0.518
7 1.46 3.60 3.16 1.000 1.218 1.471 1.328 1.420 1.376 0.496
6 1.65 3.70 3.38 1.000 1.234 1.536 1.366 1.459 1.424 0.472
5 1.86 3.80 3.57 1.000 1.253 1.596 1.412 1.507 1.483 0.447
4 2.08 3.90 3.75 1.000 1.279 1.667 1.469 1.566 1.550 0.420
3 2.31 4.00 3.91 1.000 1.312 1.754 1.543 1.643 1.614 0.391
2 2.56 4.10 4.04 1.000 1.358 1.864 1.643 1.749 1.713 0.359
1 2.82 4.20 4.16 1.000 1.426 2.013 1.787 1.903 1.848 0.325
Note: 1. y
NA
= distance from neutral axis based on linear-elastic stress distribution to mid-depth.
y
EP
= distance from neutral axis based on elasto-plastic stress distribution to mid-depth.
y
RO
= distance from neutral axis based on Ramberg-Osgood stress distribution to mid-
depth.
2. All dimensions are in inches.
130
Table 6.26 Dome-strut sections for Figure 4.15e and Figure 4.16 (Figure 6.4f)
(a) Series 1
M
AA-Y-MMCA
M
NSA-Y-TMCA
M
NSA-U-TMCA
M
NSA-U-TMCA2
M
FEM
t
f
= t
s
t
w
t
w2
d
w
M
y
M
y
M
y
M
y
M
y
y
cr
F
=
F
0.500 0.313 0.313 0.000 1.000 1.035 1.142 1.357 1.417 0.326
0.333 0.208 0.208 0.000 1.000 1.032 1.134 1.325 1.346 0.465
0.250 0.156 0.156 0.000 1.000 1.031 1.129 1.243 1.280 0.607
0.200 0.125 0.125 0.000 0.961 1.030 1.112 1.167 1.210 0.746
0.167 0.104 0.104 0.000 0.791 1.029 1.095 1.105 1.124 0.882
0.143 0.089 0.089 0.000 0.672 1.018 1.079 1.038 1.054 1.015
0.125 0.078 0.078 0.000 0.585 0.971 0.971 0.957 1.015 1.144
0.111 0.069 0.069 0.000 0.517 0.872 0.872 0.871 0.983 1.270
0.100 0.063 0.063 0.000 0.464 0.793 0.793 0.793 0.956 1.393
0.091 0.057 0.057 0.000 0.420 0.728 0.728 0.728 0.929 1.512
(b) Series 2
M
AA-Y-MMCA
M
NSA-Y-TMCA
M
NSA-U-TMCA
M
NSA-U-TMCA2
M
FEM
t
f
= t
s
t
w
t
w2
d
w
M
y
M
y
M
y
M
y
M
y
y
cr
F
=
F
0.500 0.313 0.938 0.375 1.000 1.059 1.171 1.382 1.481 0.282
0.333 0.208 0.625 0.458 1.000 1.058 1.165 1.356 1.382 0.402
0.250 0.156 0.469 0.500 1.000 1.057 1.161 1.311 1.321 0.524
0.200 0.125 0.375 0.525 0.990 1.056 1.146 1.240 1.261 0.645
0.167 0.104 0.313 0.542 0.808 1.055 1.120 1.171 1.194 0.763
0.143 0.089 0.268 0.554 0.682 1.054 1.094 1.111 1.123 0.879
0.125 0.078 0.234 0.563 0.590 1.033 1.069 1.051 1.061 0.991
0.111 0.069 0.208 0.569 0.520 0.991 0.991 0.975 1.020 1.100
0.100 0.063 0.188 0.575 0.464 0.900 0.900 0.899 0.992 1.207
0.091 0.057 0.171 0.580 0.420 0.827 0.827 0.827 0.967 1.311
Note: All dimensions are in inches.
131
6.5 Uniaxial Tensile Coupon Test Results
Table 6.27 Uniaxial tensile coupon test results (a), (b) uniform I-sections (c)
tapered I-section (d) mullion section
(a)
specimen name F
y
(ksi)
y
F
u
(ksi)
u
(ksi) E (ksi)
1t1 (top flange) 29.91 0.00497 34.24 0.06250 10068
1t2 (top flange) N.A. N.A. N.A. N.A. N.A.
1w1 (web) 29.82 0.00514 33.78 0.05400 9509
1w2 (web) 29.92 0.00506 33.89 0.06580 9775
1b1 (bottom flange) 28.84 0.00508 33.41 0.06270 9359
1b2 (bottom flange) 29.57 0.00598 33.98 0.06700 7430
average 29.61 0.00525 33.86 0.06240 9228
c.o.v. 0.015 0.079 0.009 0.081 0.113
6063-T6 (average in AA, 2000c) 31.00 0.00510 35.00 N.A. 10000
6063-T6 (minimum in AA, 2000c) 25.00 0.00448 30.00 N.A. 10100
(b)
specimen name F
y
(ksi)
y
F
u
(ksi)
u
(ksi) E (ksi)
2t1 (top flange) 30.26 0.00519 35.09 0.06294 9498
2t2 (top flange) 30.38 0.00527 35.12 0.06407 9289
2w1 (web) 30.72 0.00523 35.38 0.06810 9526
2w2 (web) 31.02 0.00538 35.52 0.06690 9189
2b1 (bottom flange) 30.68 0.00520 35.50 0.06400 9596
2b2 (bottom flange) 30.25 0.00509 35.10 0.06750 9786
average 30.55 0.00522 35.29 0.06559 9481
c.o.v. 0.010 0.019 0.006 0.033 0.023
6063-T6 (average in AA, 2000c) 31.00 0.00510 35.00 N.A. 10000
6063-T6 (minimum in AA, 2000c) 25.00 0.00448 30.00 N.A. 10100
132
Table 6.27 (Continued)
(c)
specimen name F
y
(ksi)
y
F
u
(ksi)
u
(ksi) E (ksi)
t2t1 (top flange) 41.00 0.006140 43.52 0.06830 9902
t2t2 (top flange) 40.81 0.006048 43.16 0.06783 10081
t2w1 (web) 40.58 0.006047 43.11 0.07187 10028
t2w2 (web) 40.15 0.005961 42.95 0.08132 10136
t2b1 (bottom flange) 41.45 0.006072 43.85 0.06134 10179
t2b2 (bottom flange) 41.46 0.006003 43.82 0.06960 10355
average 40.91 0.006045 43.40 0.07004 10113
c.o.v. 0.012 0.010 0.009 0.094 0.015
6061-T6 (average in AA, 2000c) 40.00 0.00600 45.00 N.A. 10000
6061-T6 (minimum in AA, 2000c) 35.00 0.00547 38.00 N.A. 10100
(d)
specimen name F
y
(ksi)
y
F
u
(ksi)
u
(ksi) E (ksi)
m2t1 (top flange) 22.66 0.004412 25.73 0.048649 9394
m2t3 (top flange) 22.62 0.004217 26.34 0.043504 10203
m2b1 (bottom flange) 21.68 0.004300 25.46 0.054104 9426
m2b2 (bottom flange) 22.25 0.004292 25.81 0.047961 9709
m2w1 (web) 23.68 0.004403 27.42 0.053167 9853
m2w2 (web) 23.27 0.004362 26.98 0.043095 9853
m2w3 (web) 23.53 0.004422 27.24 0.048226 9714
average 22.81 0.004344 26.43 0.048387 9736
c.o.v 0.032 0.018 0.030 0.088 0.028
6063-T5 (average in AA, 2000c) 21.00 0.00410 27.00 N.A. 10000
6063-T5 (minimum in AA, 2000c) 16.00 0.00358 22.00 N.A. 10100
Note: a. Gage length used in these tests is one inch.
b.
y
= strain at the yield stress =
y
/E + 0.002.
u
= strain at the ultimate stress.
c. Results from test specimens of 1t1, 1t2, 1w2, and 1b2 are based on extensometer
measurements only. All others are based on both extensometer and strain gage measurements
133
6.6 Local Geometric Initial Imperfection Measurements
Before the tests have been conducted, initial geometric imperfections were
measured on the flanges and webs of the specimens. On the flanges, the measured
imperfection data are rotated along the web-flange junctions so that the average
geometric line at the junction lies on the zero imperfection plane. On the web, the
measured data are rotated along the mid-depth. The processed imperfection data are
shown in Figure 6.5 and Figure 6.6. The maximum imperfection, which is expressed
by a larger solid circle with a corresponding value in each figure, is less than the
standard flatness tolerance by AA (2000c), except at the flange of the mullion section.
-5 0 5 10 15 20 25 30 35
0
0.5
1
1.5
-60
-50
-50
-50
-50 -50 -50
-40
-40
-30
-30
-30
-20
-20
-10
-10
0 0
10
imperfection factor = x 1"/10000
imperfection limitation by ASD = 40"/10000
-68.67
c
r
o
s
s
s
e
c
t
i
o
n
a
l
l
o
c
a
t
i
o
n
(
i
n
.
)
(a)
-5 0 5 10 15 20 25 30 35
0
1
2
3
4
5
6
-10
-5
-5
-5 -5
-5
-5 0
0
0
0
5
5
5
5
5
10
10
10
10
15
15
20
imperfection factor = x 1"/10000
imperfection limitation by ASD = 376"/10000
24.33
c
r
o
s
s
s
e
c
t
i
o
n
a
l
l
o
c
a
t
i
o
n
(
i
n
.
)
(b)
Note: All dimensions are in inches.
Figure 6.5 Imperfection measurements of a mullion section (a) top flange (b) web
longitudinal location
longitudinal location
134
10 15 20 25 30 35 40 45 50
0
0.5
1
1.5
2
2.5
-15
-10
-10
-10
-5
-5
-5
-5
0
0 0
0
0
0
0
0
5
5
5
5
5
5
5
5
5
longitudinal location (in.)
c
r
o
s
s
s
e
c
t
i
o
n
a
l
l
o
c
a
t
i
o
n
(
i
n
.
)
imperfection factor = x 1"/10000
tolerance by AA (2000c) = 50"/10000
top flange
-18.0765
(a)
10 15 20 25 30 35 40 45 50
0
0.5
1
1.5
2
2.5
-15
-10
-10
-10
-10
-5
-5
-5
0
0
0 0
5
10
longitudinal location (in.)
c
r
o
s
s
s
e
c
t
i
o
n
a
l
l
o
c
a
t
i
o
n
(
i
n
.
)
imperfection factor = x 1"/10000
tolerance by AA (2000c) = 50"/10000
bottom flange
-18.0663
(b)
10 15 20 25 30 35 40 45 50
0.5
1
1.5
2
-4
-4
-4
-4
-3 -3
-3
-3
-3
-3
-3
-3
-3
-3
-3
-3
-2 -2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-1
-1
-1
-1
-1
-1 -1 -1
-1
-1
-1
0
0
0
0
0
0
0
0
0
0
0
0
1
1
1
1
1
1
1 1
1
1
1
1
1
1
1 1
2
2
2
2
2
2 2
2
2
2
2
2 3
3
3
3
3 3
3
3
3
3
3
3
3
4
4
4
4
4
4
longitudinal location (in.)
c
r
o
s
s
s
e
c
t
i
o
n
a
l
l
o
c
a
t
i
o
n
(
i
n
.
)
imperfection factor = x 1"/10000
tolerance by AA (2000c) = 112"/10000
web
4.9334
(c)
Note: All dimensions are in inches.
Figure 6.6 Imperfection measurements of an I-3x1.64 section (a) top flange (b)
bottom flange (c) web
135
BIBLIOGRAPHY
The Aluminum Association, AA (2000a). The Specification for Aluminum Structures.
The Aluminum Association.
The Aluminum Association, AA (2000b). The Aluminum Design Manual. The
Aluminum Association.
The Aluminum Association, AA (2000c). Aluminum Standards and Data. The
Aluminum Association.
Aluminum Company of America, Alcoa (1958). Alcoa Structural Handbook, A
Design Manual for Aluminum. Aluminum Company of America.
American Institute of Steel Construction, AISC (1998). Manual of Steel Construction
Load and Resistance Factor Design. American Institute of Steel Construction.
American Iron and Steel Institute, AISI (1996). Cold-Formed Steel Design Manual.
American Iron and Steel Institute.
Anderson, R.A., Anderson M.S. (1956). Correlation of Crippling Strength of Plate
Structures with Material Properties. NACA Technical note 3600, Langley
Memorial Aeronautical Laboratory.
Bleich, F. (1952). Buckling Strength of Metal Structures, McGraw-Hill.
Chehil, D.S. and Dua, S.S. (1973). Bucking of rectangular plates with general
variation in thickness. Journal of Applied Mechanics Transaction, ASME, 40(3),
745-751.
Cheung, Y.K., (1976). Finite strip method in structural analysis. Pergamon Press,
New York
Clark, J.W., Rolf, R.L. (1966). Buckling of Aluminum Columns, Plates, and Beams.
Journal of the Structural Division, ASCE, Vol. 92, Proc. Paper 4838.
De Matteis, G., Moen, L.A., Langseth, M., Landolfo, R., Hopperstad, O.S., and
Mazzolani, F.M. (2001). Cross-Sectional Classification for Aluminum Beams-
Parametric Study. Journal of the Structural Engineering, ASCE, Vol. 127, No. 3,
Paper No. 22238.
Department of Defense, the United States of America, DOD (1994). Military
Handbook Volume 1 and 2; Metallic Materials and Elements for Aerospace
Vehicle Structures. Department of Defense, the United States of America.
135
136
Eberwien, U., Valtinat, G. (2001). The fullness method: A direct procedure for
calculation of the bending moment of a symmetrical aluminum cross section. The
8th International Conference in Aluminum (INALCO), Munich, Germany, March.
European Committee for Standardization. Eurocode 9: Design of Aluminium
Structures. Editorial Panel Version 2 (1996), European Committee for
Standardization, December.
Faella, C., Mazzolani, F.M., Piluso, V., and Rizzano, G. (2000). Local Buckling of
Aluminum Members: Testing and Classification. Journal of the Structural
Engineering, ASCE, Vol. 126, No. 3, Paper No. 20270.
Gaylord, Jr., E.H., Gaylord, C.N. (1979). Structural Engineering Handbook. McGraw-
Hill.
Gerard, G. and Becker, H. (1957). Handbook of Structural Stability. Part I- Buckling
of Flat Plates. NACA Technical Note 3781. National Advisory Committee for
Aeronautics.
Hibbitt, Karlsson & Sorensen, Inc. (1998). ABAQUS Version 5.8, Hibbitt, Karlsson &
Sorensen, Inc.
Hill, H.N., Clark J.W. (1955). Straight-line Column Formulas for Aluminum Alloys.
Alcoa Technical Paper No.12, Pittsburgh, Pennsylvania.
Huebner, K.H., Thornton, E.A., Byrom, T.G. (1995). The Finite Element Method for
Engineers. 3
rd
Ed, John Wiley & Sons, Inc.
Jombock, J.R., Clark, J.W. (1968). Bending Strength of Aluminum Formed Sheet
Members. Journal of the Structural Division, ASCE, Vol. 94, No. ST2, Proc.
Paper 5816.
Kim, Y. (2000), Behavior and Design of Laterally Supported Doubly Symmetric I-
Shaped Extruded Aluminum Sections. M.S. Thesis, Cornell University, Ithaca,
New York.
Kissell, J.R., Ferry, R.L. (1995). Aluminum Structures, John Wiley & Sons, Inc.
Kobayashi, H., Sonoda, K., (1990) "Buckling of rectangular plates with tapered
thickness." Journal of Structural Engineering, ASCE, 116(5), 1278-1289.
Mazzolani, F.M. (1985). Aluminum Alloy Structures. 1
st
Ed., Pitman Publishing Inc.
Mazzolani, F.M. (1995). Aluminum Alloy Structures. 2
nd
Ed., E & FN Spon.
Mazzolani, F.M. and Piluso, V. (1997). Prediction of the Rotational Capacity of
Aluminum Alloy Beams. Thin-Walled Structures, Vol. 27, No.1, pp.103-116.
137
McGuire, W., Gallagher, R.H., Ziemian, R.D. (2000). Matrix Structural Analysis. 2nd
Ed. John Wiley and Sons, Inc.
Mennink, J. (2002). Cross-Sectional Stability of Aluminum Extrusions. Ph.D.
Thesis. Eindhoven University of Technology, The Netherlands.
Mizusawa, T., (1993), Buckling of rectangular Mindlin plates with tapered thickness
by the spline strip method. International Journal of Solids Structures, 30(12),
1663-1677.
Ohga, M., Shigematsu, T., Kawaguchi, K., (1995) Buckling analysis of thin-walled
members with variable thickness. Journal of Structural Engineering, ASCE,
121(6), 919-924.
Pines, S., and Gerard, G. (1947). Instability analysis and design of an efficiently
tapered plate under compressive loading. Journal of the Aeronautical Sciences,
14(10), 594-599.
Plecher, R. (2000). Aluminum for Bridges, Evaluation of Existing Structures in
North America. Ph.D. Thesis. Technical University of Munich, Munich,
Germany.
Ramberg, W., Osgood, W.R. (1943) Description of Stress-Strain Curves by Three
Parameters. NACA Technical Note 902, National Bureau of Standards.
Reck, P., Pekz, T., and Winter, G. (1975). Inelastic strength of cold-formed steel
beams, Journal of Structural Division, ASCE, 101 (ST11), Nov. 1975.
Schafer, B.W. (1997). Cold-formed Steel Behavior and Design: Analytical and
Numerical Modeling of Elements and Members with Longitudinal Stiffeners.
Ph.D. Thesis, Cornell University, Ithaca, New York.
Schafer, B.W. and Pekz, T. (1998). Direct Strength Prediction of Cold-Formed Steel
Members Using Numerical Elastic Buckling Solutions. Fourteenth International
Specialty Conference on Cold-Formed Steel Structures, St. Louis, Missouri.
October.
Sharp, M. L. (1966). Longitudinal Stiffeners for Compression Members. Journal of
the Structural Division, ASCE, Vol. 92, No. ST5, Proc. Paper No. 4948.
Sharp, M.L. (1993). Behavior and Design of Aluminum Structures, McGraw-Hill.
Sooi, T. K. and Pekz, T. (1993). Behavior of Component Elements of Aluminum
Members. Research Report No.93-1, School of Civil and Environmental
Engineering, Cornell University, Ithaca, New York.
138
Stowell E.Z. (1948). A Unified Theory of Plastic Buckling of Columns and Plates.
NACA Technical Note 1556, Langley Memorial Aeronautical Laboratory.
Stowell E.Z. (1950). Compressive Strength of Flanges. NACA Technical Note 2020,
Langley Memorial Aeronautical Laboratory.
Templin R.L., Sturm R.G., Hartmann E.C., Holt M. (1938). Column Strength of
Various Aluminum Alloys. Alcoa Technical Paper No.1, Pittsburgh,
Pennsylvania.
Wittrick, W.H., and Ellen, C.H. (1962). Buckling of tapered rectangular plates in
compression. The Aeronautics Quarterly, 13(4), 308-326.